Oscillatory Coupling between Basal Ganglia, Cortex and ... · This thesis investigated local and...
Transcript of Oscillatory Coupling between Basal Ganglia, Cortex and ... · This thesis investigated local and...
Oscillatory Coupling between Basal Ganglia,
Cortex and Muscle in Parkinson’s Disease
Inaugural-Dissertation
zur Erlangung des Doktorgrades
der Mathematisch-Naturwissenschaftlichen Fakultät
der Heinrich-Heine-Universität Düsseldorf
vorgelegt von
Jan Hirschmann
aus Göttingen
Düsseldorf, Februar 2014
Aus dem Institut für Klinische Neurowissenschaften
und Medizinische Psychologie
der Heinrich-Heine-Universität Düsseldorf
Gedruckt mit der Genehmigung der
Mathematisch-Naturwissenschaftlichen Fakultät der
Heinrich-Heine-Universität Düsseldorf
Referent: Prof. Dr. Alfons Schnitzler
Korreferent: Prof. Dr. Tobias Kalenscher
Tag der mündlichen Prüfung: 10.12.2013
TABLE OF CONTENTS
GLOSSARY ........................................................................................................................................... 5
ZUSAMMENFASSUNG ........................................................................................................................ 6
SUMMARY ........................................................................................................................................... 8
1 INTRODUCTION ....................................................................................................................... 11
1.1 Non-invasive measurement of neuronal oscillations ............................................... 11
1.2 Analysis of neuronal oscillations .................................................................................. 12
1.2.1 Spectral analysis ......................................................................................................... 12
1.2.2 Source reconstruction ................................................................................................ 13
1.3 The basal ganglia ............................................................................................................. 14
1.3.1 Anatomy ...................................................................................................................... 14
1.3.2 Function ....................................................................................................................... 17
1.4 Parkinson’s disease ......................................................................................................... 17
1.4.1 Symptoms .................................................................................................................... 18
1.4.2 Pathogenesis ............................................................................................................... 18
1.4.3 Pathophysiology according to the classical rate model........................................... 19
1.4.4 Treatment .................................................................................................................... 20
1.5 Oscillations in Parkinson’s disease .............................................................................. 22
1.5.1 Alpha oscillations ........................................................................................................ 22
1.5.2 Beta oscillations .......................................................................................................... 23
1.5.3 Gamma and high frequency oscillations ................................................................... 23
2 AIMS ........................................................................................................................................... 25
3 PARADIGM ................................................................................................................................ 26
4 STUDY 1: Distinct oscillatory STN-cortical loops revealed by simultaneous MEG
and local field potential recordings in patients with Parkinson's disease ..................... 27
4.1 Methods ............................................................................................................................. 28
4.2 Results ................................................................................................................................ 28
4.2.1 Distribution of STN-cortical coherence across brain areas .................................... 28
4.2.2 Distribution of STN-STG and STN-M1 coherence across electrode contacts ........ 29
4.3 Discussion ......................................................................................................................... 29
4.4 Conclusions ....................................................................................................................... 30
5 STUDY 2: Differential modulation of STN-cortical and cortico-muscular
coherence by movement and levodopa in Parkinson's disease ........................................ 30
5.1 Methods ............................................................................................................................. 31
5.2 Results ................................................................................................................................ 32
5.2.1 Effects of movement and medication ........................................................................ 32
5.2.2 Correlation with clinical parameters ........................................................................ 32
5.3 Discussion ......................................................................................................................... 32
5.4 Conclusions ....................................................................................................................... 34
6 STUDY 3: A direct relationship between oscillatory STN-cortex coupling and rest
tremor in Parkinson’s disease ................................................................................................. 35
6.1 Methods ............................................................................................................................. 36
6.2 Results ................................................................................................................................ 36
6.2.1 Sensor level ................................................................................................................. 36
6.2.2 Source level ................................................................................................................. 36
6.3 Discussion ......................................................................................................................... 37
6.4 Conclusions ....................................................................................................................... 38
7 GENERAL DISCUSSION ............................................................................................................ 38
8 OUTLOOK .................................................................................................................................. 39
9 REFERENCES ............................................................................................................................. 41
10 ERKLÄRUNG .............................................................................................................................. 48
11 DANKSAGUNG ........................................................................................................................... 49
12 APPENDIX ................................................................................................................................. 50
5 5Glossary
Glossary
EEG electroencephalography
MEG magnetoencephalography
SQUID superconductive quantum interference device
LCMV linear constraint minimum variance
DICS dynamic imaging of coherent sources
M1 primary motor cortex
GPe external segment of the globus pallidus
GPi internal segment of the globus pallidus
STN subthalamic nucleus
SNc substantia nigra pars compacta
SNr substantia nigra pars reticulata
GABA γ-aminobutyric acid
PD Parkinson’s disease
MPTP 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine
6-OHDA 6-hydroxydopamine
L-DOPA L-3,4-dihydroxyphenylalanine; also known as levodopa
DBS deep brain stimulation
LFP local field potential
EMG electromyography
UPDRS unified Parkinson’s disease rating scale
STG superior temporal gyrus
ROI region of interest
PMC premotor cortex
PPC posterior parietal cortex
6 6Zusammenfassung
Zusammenfassung
Morbus Parkinson ist eine progressive, neurodegenerative Erkrankung des zentralen
Nervensystems, deren Symptome unter anderem mit der tiefen Hirnstimulation
behandelt werden. Die dafür notwendigen chirurgischen Eingriffe ermöglichen es,
Hirnaktivität der Basalganglien in Form lokaler Feldpotentiale (LFPs) aufzuzeichnen.
Zahlreiche Studien, die sich mit der Analyse von LFPs beschäftigt haben, lieferten
Hinweise darauf, dass neuronale, oszillatorische Aktivität bei Parkinson-Patienten
pathologisch verändert ist. Zudem konnte gezeigt werden, dass neuronale Oszillationen
nicht nur lokal sondern auch interregional synchronisiert sind. So sind beispielsweise
Beta Oszillationen (13-35 Hz) im Nucleus subthalamicus (STN) und Cortex kohärent.
Diese Doktorarbeit befasste sich mit lokaler und interregionaler neuronaler
Synchronisation bei Parkinson-Patienten. Im Mittelpunkt stand die Charakterisierung
der oszillatorischen Kopplung zwischen STN, Cortex und Muskel.
Basierend auf einem einheitlichen Paradigma wurden drei Studien durchgeführt. In
allen Studien wurden LFP Messungen im STN, Magnetenzephalographie (MEG) und
Elektromyographie simultan durchgeführt. Akinetisch-rigide (Studien 1 und 2) und
tremor-dominante Parkinson-Patienten (Studie 3) wurden in Ruhe, während der
Ausführung einer Halteaufgabe (Elevation des Unterarms) und während der Ausführung
einer Bewegungsaufgabe (repetitives Öffnen und Schließen der Faust) untersucht. Die
Messungen wurden zunächst nach Entzug dopaminerger Medikation durchgeführt und
nach Verabreichung von Levodopa wiederholt. Die Quantifizierung von lokaler
Synchronisation erfolgte durch Berechnung von Power. Die Quantifizierung von
interregionaler oszillatorischer Kopplung erfolgte durch Berechnung von Kohärenz.
Studie 1 beschäftigte sich mit der räumlichen Verteilung von STN-cortikaler Kohärenz
im Ruhezustand. Diese erwies sich als frequenzabhängig. Im Alpha-Band (8-12 Hz) war
der STN vornehmlich an temporale Areale gekoppelt, insbesondere an den Gyrus
temporalis superior (STG). Im Beta-Band hingegen zeigten der primäre motorische
(M1), der primäre somatosensorische und der prämotorische Cortex die stärkste
Kohärenz. Räumliche Kohärenzmaxima waren nahezu ausschließlich ipsilateral zum
STN lokalisiert. Eine nähere Betrachtung der einzelnen Elektrodenkontakte im STN
ergab, dass die Alpha-Kopplung zum STG an allen Kontakten zu beobachten war,
7 7Zusammenfassung
während die Beta-Kopplung zu M1 auf einen oder zwei Kontakte beschränkt war. Diese
Ergebnisse zeigen, dass nur ein eingegrenzter Bereich innerhalb oder im Umfeld des
STN synchron mit M1 oszilliert. Die Organisation der anatomischen Verbindungen
zwischen M1 und STN lässt vermuten, dass es sich hierbei um den dorsolateralen Teil
des STN handelt. Neben der Charakterisierung der frequenzabhängigen Verteilung von
STN-cortikaler Kohärenz demonstrierte Studie 1 die Möglichkeit, oszillatorische
Kopplungen mittels simultaner LFP-MEG Messungen zu untersuchen.
Studie 2 befasste sich mit der Modulation von STN-cortikaler und cortiko-muskulärer
Kohärenz durch Bewegung und Medikation. Es wurde gezeigt, dass Alpha- und Beta-
Kohärenz zwischen M1 und dem Extensor-Muskel des Unterarms durch Bewegung im
Vergleich zur Halteaufgabe reduziert wird. Levodopa zeigte keine Wirkung auf die
cortiko-muskuläre Kohärenz. Allerdings führte die Medikation zu einer Reduktion der
Beta-Kohärenz zwischen M1 und STN. Diese korrelierte wider Erwarten nicht mit der
klinischen Verbesserung der Beweglichkeit. Stattdessen wurde eine negative
Korrelation zwischen Beta-Kohärenz und Unterbeweglichkeit festgestellt, d.h. Patienten
mit einer stärkeren Kohärenz waren besser beweglich als Patienten mit einer
schwächeren Kohärenz. Studie 2 zeigt, dass STN-cortikale und cortiko-muskuläre
Kohärenz unabhängig voneinander moduliert werden können. Die negative Korrelation
zwischen Beta-Kohärenz und Unterbeweglichkeit lässt Zweifel an der weit verbreiteten
Annahme aufkommen, dass starke Beta-Kohärenz zwischen STN und M1 einen
pathologischen Mechanismus der Parkinson-Erkrankung darstellt.
Ziel von Studie 3 war es, mögliche Veränderungen der STN-cortikalen und cortico-
muskulären Kohärenz zu ermitteln, die mit dem Einsetzen des Parkinson-typischen
Ruhetremors einhergehen. Die Ergebnisse zeigen, dass die neuronale Synchronisation in
der Tremor-Frequenz und der doppelten Tremor-Frequenz ansteigt sobald der Tremor
auftritt. Ein Anstieg konnte für Power im STN sowie für die oszillatorische Kopplung
zwischen STN, Cortex und Muskel nachgewiesen werden. Eine Analyse auf Quell-Ebene
offenbarte, dass M1, der prämotorische Cortex sowie der posteriore Parietallappen
während des Tremors eine erhöhte Kohärenz mit den Unterarmmuskeln aufweisen und
zudem untereinander synchronisiert sind. Studie 3 belegt, dass oszillatorische Kopplung
in der Tremor-Frequenz ein neuronales Korrelat des Ruhetremors ist.
8 8Summary
Die im Rahmen der Dissertation durchgeführten Studien zeigen, dass oszillatorische
Kopplung zwischen STN, Cortex und Muskel durch Bewegung, dopaminerge Medikation
und Tremor moduliert wird und mit Beweglichkeit korreliert. Diese Ergebnisse
verdeutlichen die wichtige Rolle von synchronen Oszillationen in der Pathophysiologie
von Morbus Parkinson und ermöglichen eine Zuordnung von
Synchronisationsprozessen und Symptomen, die möglicherweise für eine gezielte,
therapeutische Manipulation von pathologischen Oszillationen relevant werden könnte.
Summary
Parkinson’s disease (PD) is a progressive, neurodegenerative disease of the central
nervous system which is treated, amongst other therapeutic interventions, by deep
brain stimulation (DBS). The surgical procedure for DBS provides the unique
opportunity to record local field potentials (LFPs) from the human basal ganglia.
Numerous studies investigating LFPs found indications for pathological alterations of
synchronous oscillations in PD. These studies also showed that synchronization occurs
locally as well as between distant brain regions. For example, it was demonstrated that
LFPs recorded from the subthalamic nucleus (STN) are coherent with cortical
oscillations in the beta band (13- 35 Hz).
This thesis investigated local and interregional synchrony in PD. Its major aim was to
characterize oscillatory coupling between STN, cortex and muscle.
Three studies were performed which were all based on the same experimental
paradigm. In all studies, STN LFPs, the magnetoencephalogram (MEG) and the
electromyogram of forearm muscles were recorded simultaneously. Akinetic-rigid
(studies 1 and 2) and tremor-dominant patients (study 3) were recorded at rest, during
a static motor task (forearm elevation) and during repetitive movement (opening and
closing of the fist). Measurements took place after withdrawal of dopaminergic
medication and were repeated following administration of levodopa. Local synchrony
was quantified by power and interregional oscillatory coupling was quantified by
coherence.
Study 1 investigated the spatial distribution of STN-cortical coherence at rest.
Interestingly, the distribution was found to be frequency-dependent. STN alpha (8-12
9 9Summary
Hz) oscillations were predominantly coherent with oscillations in temporal areas. In
particular, there was strong alpha coherence with superior temporal gyrus (STG). In the
beta band, however, coherence was strongest with primary motor cortex (M1), primary
somatosensory cortex and premotor cortex. The vast majority of spatial coherence
maxima were located ipsilateral to the STN. Inspection of the distribution of coherence
across STN electrode contacts revealed that alpha band coupling to STG was distributed
homogenously across contacts. In contrast, beta band coupling to M1 was usually
restricted to one or two contacts. The results suggest that beta synchrony with M1 is
confined to a circumscribed area within or near the STN. Given the organization of
anatomical connections between M1 and STN, it is reasonable to assume that beta band
coupling is restricted to the dorsolateral portion of the STN. Apart from characterizing
the frequency-dependent distribution of STN-cortical coherence, study 1 demonstrated
that simultaneous LFP-MEG recordings are a powerful tool for studying oscillatory
coupling in PD.
Study 2 assessed modulations of STN-cortical and cortico-muscular coherence by
movement and dopaminergic medication. It showed that coherence between M1 and the
forearm extensor muscle is reduced by repetitive movement compared to static
contraction in the alpha and beta band. Levodopa did not affect cortico-muscular
coherence but led to a reduction of beta coherence between M1 and STN. Surprisingly,
this reduction did not correlate with the improvement of motor symptoms. However,
there was a negative correlation between beta coherence and akinesia in the OFF state,
i.e. patients with strong coherence showed better motor performance than patients with
weak coherence. Study 2 demonstrates that STN-cortical and cortico-muscular
coherence can be modulated independently. The negative correlation between beta
coherence and akinesia challenges the widespread belief that strong STN-cortical beta
coherence reflects a pathological mechanism.
Study 3 aimed at characterizing the changes in STN-cortical and cortico-muscular
coherence associated with the manifestation of parkinsonian rest tremor. It was found
that neuronal synchrony at tremor frequency and double tremor frequency increases
when tremor emerges. Increases were observed for STN power and oscillatory coupling
between STN, cortex and muscle. Source level analysis revealed that M1, premotor
cortex and posterior parietal cortex show increased coherence with forearm muscles
during tremor and are synchronized with each other at tremor frequency and its first
10 10Summary
upper harmonic. Study 3 demonstrates that oscillatory coupling at tremor frequency is a
genuine neural correlate of rest tremor in PD.
The presented studies reveal modulations of oscillatory coupling by movement,
dopaminergic medication and tremor. Further, they demonstrate that coupling is
correlated with motor performance. In summary, the results emphasize the pivotal role
of synchronous oscillations in PD pathophysiology and provide insights into the
association between neuronal synchronization and PD symptoms. These insights might
become relevant for therapeutic manipulation of pathological oscillatory processes.
11 11Introduction
1 Introduction
Neuronal oscillations are periodic fluctuations of membrane or extracellular potentials
which can be measured at several scales. Ever since Hans Berger recorded oscillations
non-invasively in humans for the first time (Berger, 1929), they attracted attention in
the scientific community. Today, it is clear that neuronal oscillations are directly related
to behavior (Buzsaki, 2006). Moreover, there is good evidence that they are
pathologically altered in several neurological disorders. In particular, altered oscillatory
activity was found to be a hallmark of Parkinson’s disease (PD). The following
introduction will provide an overview of how neuronal oscillations are measured and
analyzed, describe PD on the symptom and the neurophysiological level and will finally
present the current knowledge on the intricate relationship between neuronal
oscillations and PD symptoms.
1.1 Non-invasive measurement of neuronal oscillations
One of the main reasons for the tremendous interest in neuronal oscillations is that they
can be measured non-invasively in humans. Therefore, they can be related to complex
behaviors. Traditionally, studies on neuronal oscillations divide the spectrum into five
frequency bands: theta (<4 Hz), delta (4-7 Hz), alpha (8-12 Hz), beta (13-30 Hz) and
gamma (>30 Hz). Importantly, these different frequency bands have been associated
with different perceptual, motor and cognitive functions. Non-invasive recordings of
oscillatory activity can be performed using electroencephalography (EEG) or
magnetoencephalography (MEG).
In EEG, surface electrodes are attached to the skull to monitor changes in electric
potentials over time. MEG is similar in principle but measures the magnetic rather than
the electric field (reviewed in Hämäläinen et al., 1993). Since the magnetic fields
resulting from brain activity are very small, their detection requires extremely sensitive
sensors called superconductive quantum interference devices (SQUIDs). The small
amplitude of brain signals further implies that they are masked by ambient fields such
as the earth’s steady field or the fields produced by muscle contraction. In fact, the latter
surmount brain magnetic fields by several orders of magnitude. Thus, MEG is conducted
within a shielded room that blocks external magnetic fields by virtue of its material (mu-
metal) and by active cancellation in custom-made circuits. In addition, special types of
12 12Introduction
sensors are used to minimize interference. So-called gradiometers are designed to
measure spatial field gradients rather than the fields per se, leading to decreased
sensitivity to distant sources. Finally, several online and offline processing tools are
available to suppress interference.
Both EEG and MEG measure spatial sums of signals produced in the brain such as
synaptic currents, action potentials, calcium spikes or intrinsic membrane responses
(reviewed in Buzsáki et al., 2012). Detection by EEG and MEG requires that several
thousand events occur simultaneously. Moreover, individual contributions must not
cancel. Whether cancellation occurs depends on timing and geometry. For example,
radial currents produce almost no detectable signal in MEG due to magnetic field
cancellation. Thus, MEG is mainly sensitive to tangential sources.
1.2 Analysis of neuronal oscillations
Oscillatory brain activity measured by MEG is typically first analyzed on the sensor level
in order to identify the frequencies and/or time points relevant to the experimental task.
However, these may also be defined a priori. Following frequency and/or epoch
selection, activity can be localized in order to identify the brain areas that give rise to the
effects observed on the sensor level.
1.2.1 Spectral analysis
Since neuronal oscillations are periodic, they can be decomposed into a sum of sine
waves by means of Fourier transformation which yields phase and amplitude as a
function of frequency. As electrophysiological recordings are finite, the latter are
estimates rather than the true Fourier coefficients of the underlying process. A range of
interesting parameters can be derived from these coefficients.
Spectral power is by far the most frequently analyzed parameter. It is defined as the
squared amplitude at a given frequency, i.e. it quantifies signal energy. Changes in power
across experimental conditions are usually interpreted as changes in the degree of
synchronization within the recorded population of neurons, but could also indicate a
change in the number of neurons contributing to the signal.
13 13Introduction
Coherence is a measure often employed to quantify the level of synchrony between two
signals. It reflects the degree of amplitude and phase co-variation across data segments
or trials. In other words, it measures whether two signals tend to keep a fixed phase
difference and a fixed amplitude product over time. Coherence values range between
zero (independent) and one (completely synchronous). Since a value of zero requires a
perfectly symmetric distribution of individual phase differences, it is never measured in
practice. Thus, coherence is biased towards non-zero values and the bias increases as
the amount of available data segments decreases (Maris et al., 2007). Therefore, a
comparison of coherence across experimental conditions is only sensible if the number
of data segments is approximately equal for all conditions.
Coherence between signals from two distant brain areas indicates that rhythmic activity
in these areas is coordinated in time. Therefore, coherence is often considered a
measure of functional connectivity (reviewed in Varela et al., 2001). It was proposed
that a neuronal population A may regulate its impact on a second population B by timing
its input relative to intrinsic membrane potential oscillations in B (Fries, 2005). In this
context, the term “impact” describes the ability of A to trigger action potentials in B. For
example, the impact would be maximal if A’s input arrives when B happens to be
maximally depolarized. Such coordinated, rhythmic input would result in high
coherence.
1.2.2 Source reconstruction
A fundamental challenge in the analysis of EEG and MEG signals is to identify the sources
of activity. This challenge is usually referred to as the “inverse problem”. To solve the
inverse problem, one typically first solves the so-called “forward problem”. The forward
problem is solved for each location of interest separately. For location j, the solution is a
model that quantifies what sensor measurement would result from a current of unit
amplitude at location j. The solution to the inverse problem is the inverse of the forward
model, i.e. it maps sensor measurements to source activity. Due to the enormous
number of possible sources and the limited number of sensors there is in principle too
little information to reconstruct the origin of a given sensor measurement. Thus, the
inverse problem can only be solved by making assumptions which guarantee the
existence of a unique solution. There a various algorithms which solve the inverse
14 14Introduction
problem by making assumptions such as minimum norm, least-squares or beamforming
approaches.
Beamforming is a source localization method which sequentially scans a predefined set
of locations to obtain point-by-point estimates of source activity (reviewed in Hillebrand
et al., 2005). The estimates are computed as weighted sums of the sensor data:
y is the activity estimate for location j, xi is the ith sensor recording, wi is the weight for
this sensor and N is the number of sensors.
The studies presented in this thesis make use of two kinds of beamformers. One of these
operates on time domain data and is known as linear constraint minimum variance
(LCMV) beamformer (Van Veen and Buckley, 1988). The other is called Dynamic
Imaging of Coherent Sources (DICS; Gross et al., 2001) and operates on frequency
domain data. The idea behind both algorithms is the same: they aim at minimizing
interference. This is achieved by choosing the weights w such that the output y is as
small as possible, with the crucial constraint that activity from location j must not be
modified in any way. In consequence, the only remaining part of the signal that can be
minimized is interfering activity from other locations. The optimal weights can be found
analytically with the help of common concepts in optimization. In this work,
beamforming was primarily used to estimate coherence between cortical areas and the
basal ganglia.
1.3 The basal ganglia
The basal ganglia are a core element of the motor system. Alterations in basal ganglia
activity patterns, especially in oscillatory activity, can be observed in several
neurological disorders. In particular, such alterations are known to occur in PD.
1.3.1 Anatomy
The term “basal ganglia” refers to a group of interconnected subcortical nuclei which
receive input from the cortex and project back to the cortex via polysynaptic pathways.
Thus, cortex and basal ganglia form a loop. One outstanding characteristic of this loop is
that the functional organization of the cortex is roughly maintained. Inputs of different
15 15Introduction
modality, e.g. from motor and limbic cortical areas, as well as inputs related to different
body parts remain anatomically segregated throughout the loop (Alexander et al., 1986).
Another striking feature of the basal ganglia-cortical loop is strong convergence within
the modality-specific sub-circuits. For example, each striatal neuron in the somatomotor
circuit receives several thousand cortical inputs (Kincaid et al., 1998). These may come
from supplementary, premotor or primary motor cortex (M1) or from somatosensory
areas.
The anatomy of the basal ganglia is illustrated schematically in Fig. 1. The basal ganglia
consist of the striatum, the external and internal segment of the globus pallidus (GPe and
GPi), the subthalamic nucleus (STN) and the compact and reticular compartment of the
substantia nigra (SNc and SNr). The striatum is the main input structure of the basal
ganglia (reviewed in Gerfen and Bolam, 2010). It receives afferents from all parts of the
cortex and from the thalamus. Input is provided to medium spiny projection neurons
which make up about 95% of all striatal neurons. Medium spiny neurons use γ-
aminobutyric acid (GABA) as neurotransmitter and have a low spontaneous firing rate.
Their activity is modulated by cholinergic striatal interneurons, serotonergic input from
the raphe nuclei and dopaminergic input from SNc.
16 16Introduction
Fig 1: The functional anatomy of the basal ganglia. A) The basal ganglia in healthy subjects. The cortex provides input to the striatum. Striatal output reaches GPi and SNr via direct projections and indirectly via GPe and STN. GPi and SNr inhibit the thalamus which projects back to the cortex. B) The basal ganglia in Parkinson’s disease according to the classical rate model (Albin et al., 1989; DeLong, 1990). Reduced activity in the direct pathway leads to decreased inhibition of GPi and SNr. Increased activity in the indirect pathway leads to increased excitation of GPi and SNr. Both result in increased inhibition of the thalamus and decreased feedback to the cortex. Red lines indicate excitatory connections whereas black lines indicate inhibitory connections. Line thickness marks relative activity. Dotted lines indicate a change due to dopamine depletion. For the direct pathway, dopamine depletion leads to a reduction of striatal output. For the indirect pathway, dopamine depletion leads to an increase of striatal output. This difference is due to expression of different dopamine receptor types.
Striatal output reaches the output nuclei of the basal ganglia, GPi and SNr, either directly
or indirectly via GPe and STN. While activity in the direct pathway inhibits the output
structures, activity in the indirect pathway exerts excitatory influence. Excitation is
achieved by inhibition of the GPe and subsequent disinhibition of the STN which is the
only glutamatergic nucleus in the loop. When disinhibited, the STN drives GPi and SNr.
The STN also receives direct cortical afferents via the so-called hyperdirect pathway
(Nambu et al., 1996).
GPi and SNr contain GABAergic neurons projecting to the thalamus which closes the
loop by providing input to cortex and striatum. Other basal ganglia output targets are
the superior colliculus and the pedunculopontine nucleus. Importantly, the ultimate
outcome of cortical input to the basal ganglia is either an increase (indirect pathway) or
a decrease (direct pathway) of thalamic inhibition, i.e. a regulation of feedback.
17 17Introduction
1.3.2 Function
Currently, the function of the basal ganglia is not fully understood. Many theories have
been proposed but none of them satisfactorily explains all of the experimental findings.
It is undisputed that the basal ganglia can strongly influence motor behavior. Abnormal
basal ganglia activity patterns are observed in several movement disorders such as
Parkinson’s disease, Huntington’s disease, dystonia or Tourette’s syndrome (reviewed in
Wichmann and Dostrovsky, 2011). Furthermore, the basal ganglia are known to be
involved in many different types of learning and in habit formation (reviewed in Ashby
et al., 2010). Other non-motor functions are known to exist but have rarely been
investigated. For example, the basal ganglia were found to be involved in emotional tone
processing (Pell and Leonard, 2003) and in processing of motivational value (Levy and
Dubois, 2006).
With regard to motor control, suggested functions include online error correction, gain
control, the retention of over-trained motor skills and action selection. Especially the
latter hypothesis is well-known and still supported by numerous researchers (Mink,
1996). However, it is seriously challenged by the finding that the earliest changes in GPi
firing rate occur at the time of the earliest agonist muscle activity, i.e. too late for action
selection (Turner and Anderson, 1997).
Recent studies in non-human primates aimed at identifying basal ganglia motor
functions by inactivating the GPi, i.e. by blocking basal ganglia output (Desmurget and
Turner, 2008, 2010). GPi inactivation resulted in slowed and undershooting movements
but did not impair reaction time, online movement correction or the execution of over-
trained sequences. In summary, current research supports the hypothesis that the basal
ganglia are involved in setting the gain of movement. However, a general consensus has
not been reached and further research is necessary.
1.4 Parkinson’s disease
PD is a progressive, neurodegenerative disorder which was first described by James
Parkinson in 1817. Due to its relatively high prevalence (1.6% of people older than 65;
Rijk et al., 1997) it has a considerable and growing impact on society. Thus,
understanding PD is one of the major challenges to modern neurological research.
18 18Introduction
1.4.1 Symptoms
The main symptoms of PD are akinesia (poverty of movement), bradykinesia (slowness
of movement), rigidity (muscle stiffness) and rest tremor (reviewed in Lang and Lozano,
1998a, 1998b). In addition, patients often develop further motor impairments such as
postural instability or gait disturbances. Non-motor symptoms such as dementia or
depression are also common, especially in late stages of the disease. PD mainly affects
the elderly and age is the only risk factor consistently identified in epidemiological
studies.
PD can be subdivided into two types: the akinetic-rigid subtype with no or little tremor
but markedly slowed movement and the tremor-dominant subtype showing strong rest
tremor but little akinesia and rigidity (reviewed in Helmich et al., 2012). The distinction
is well-established and based on both subjective classification and automated statistical
cluster analysis of clinical data (Lewis et al., 2005). Compared to the akinetic-rigid
subtype, the tremor-dominant subtype is characterized by relatively mild impairments
at disease onset and slow disease progression in the first years following diagnosis.
However, late stage symptoms such as falls occur after similar disease duration in both
subtypes, indicating that disease progression in tremor-dominant patients accelerates in
later stages (Selikhova et al., 2009).
1.4.2 Pathogenesis
Most PD symptoms are consequences of a progressive loss of dopaminergic neurons in
SNc and other midbrain nuclei. Despite decades of intensive research, the cause of cell
death remains unknown. Therefore, the large majority of patients are classified as
suffering from idiopathic PD (of unknown origin). However, some hereditary forms exist
and a number of risk genes have been identified (reviewed in Obeso et al., 2010).
One of these risk genes is SNCA which codes for the presynaptic protein α-synuclein. In
PD patients, α-synuclein forms intracellular aggregates together with other proteins.
These aggregates are called Lewy bodies. Investigation of the distribution of Lewy
bodies at different symptomatic stages revealed a progressive spread from the
brainstem towards cortical areas (Braak et al., 2003). This and other studies showed
that a number of areas other than SNc are affected in PD, some of them much earlier.
19 19Introduction
Interestingly, the earliest Lewy bodies are found in the dorsal XI/X motor nucleus of the
glossopharyngeal and vagal nerves and in the olfactory bulb, in line with the
parasympathetic and olfactory symptoms that often precede motor impairments by
several years. Thus, PD comprises predominantly, but not exclusively, motor symptoms
which emerge years after progressive cell death has begun in a number of brain areas.
There are two forms of human parkinsonism which are known to be caused by
environmental factors: delayed-onset parkinsonism following encephalitis lethargica
and intoxication with 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP). MPTP is
sometimes accidentally consumed by heroin addicts and can result in severe akinesia
and rest tremor. As monkeys show a similar response, MPTP is used to create primate
models of PD. Rodents are much less susceptible to the substance but develop motor
symptoms reminiscent of PD after cerebral injection of 6-hydroxydopamine (6-OHDA ),
a toxic dopamine receptor agonist.
1.4.3 Pathophysiology according to the classical rate model
The classical rate model of PD pathophysiology proposes an imbalance between the
direct and the indirect pathway within the basal ganglia-cortical loop (Albin et al., 1989;
DeLong, 1990). In this model, pathological alterations are exclusively caused by a lack of
dopamine in the striatum.
Dopamine depletion has opposite effects on the direct and the indirect pathway. The
difference is due to expression of different dopamine receptor types. Striatal neurons
projecting to GPi (direct pathway) predominantly express D1 dopamine receptors
(Gerfen et al., 1990). Upon dopamine binding, medium spiny neurons expressing D1
receptors increase their firing rate, resulting in disinhibition of the thalamus (see section
1.2.1). When dopamine is lacking, disinhibition decreases and the thalamus becomes
less active, leading to decreased activity in cortical motor areas and thereby to akinesia
(Fig. 1B).
Striatal neurons in the indirect pathway predominantly express D2 receptors. In
contrast to D1 receptors, activation of D2 receptors leads to inhibition. As depicted in
Fig. 1, inhibition of striatal neurons projecting to GPe results in STN inhibition. In PD,
20 20Introduction
dopamine levels are reduced so that the STN is less inhibited and therefore drives GPi. In
turn, GPi excessively inhibits the thalamus.
In summary, the classical rate model proposes that PD is characterized by too little
activity in the direct pathway and too much activity in the indirect pathway. Notably, a
recent optogenetic study in which the pathways were stimulated selectively supported
this basic hypothesis (Kravitz et al., 2010). The model explains a number of
observations. For example, activity in cortical motor areas was found to be reduced in
non-medicated PD patients compared to the medicated state (Jenkins et al., 1992).
Moreover, lesions of STN (Bergman et al., 1990) and GPi (Lozano et al., 1995) alleviate
PD motor symptoms, as predicted by the model. However, the model also has its
shortcomings. First, it is incomplete. For example, it does not incorporate the
hyperdirect pathway or projections from various subcortical nuclei back to the striatum.
Second, some of its predictions are wrong. For example, firing rate changes in GPi
following dopamine depletion are rather small, casting doubt on the hypothesis that GPi
excessively inhibits the thalamus in PD (e.g. Wichmann et al., 1999). Moreover, GPi
lesions do not produce the predicted result. According to the model, GPi lesions should
elicit dyskinesias (involuntary choreoathetotic movements) since the thalamus is not
inhibited anymore. In reality, GPi lesions abolish dyskinesias in PD patients (Laitinen et
al., 1992).
1.4.4 Treatment
To date, PD can neither be cured nor can its progression be stopped. However, its
symptoms can be treated by either pharmacological or surgical intervention. L-3,4-
dihydroxyphenylalanine, called levodopa or L-DOPA, is the most effective drug for PD
treatment. Unlike dopamine, levodopa can pass the blood-brain barrier and thus reaches
its target tissue after oral intake. In the brain, it is metabolized to dopamine and
compensates for the lack of nigral dopamine. The treatment effects are highly reliable so
that a positive response to levodopa is used as a criterion for diagnosis of PD.
While levodopa successfully restores motor capabilities, its long-term application is
associated with side effects. After several years, patients often require a higher dosage to
reach the same effect (“wearing-off”), experience unpredictable transitions between
akinetic and mobile states (“ON-OFF fluctuations”) and exhibit dyskinesias. Especially
21 21Introduction
the latter can heavily impair motor performance. Therefore, levodopa-induced
dyskinesias are a common motivation for considering alternative therapeutic options
such as deep brain stimulation (DBS).
DBS effectively alleviates PD motor symptoms and usually allows for a substantial dose
reduction of anti-parkinsonian medication (Limousin et al., 1998). In DBS, electrodes are
implanted into the target area where they deliver current pulses which are generated by
a subcutaneous stimulator. For treatment of PD, either STN or GPi are targeted.
Typically, the STN is stimulated with a frequency of 130 Hz.
While its clinical benefit is undisputed, the mechanisms of DBS are still under debate
(reviewed in Kringelbach et al., 2007). Since the outcomes of STN DBS and STN lesions
are similar, DBS is often interpreted as a “functional lesion”. The term is suggestive of an
inhibitory effect and, indeed, several studies found that DBS leads to a lasting reduction
in the firing rate of local STN neurons (e.g. Beurrier et al., 2001). However, local neurons
are not the only neural elements exposed to the electric field. Passing axons, for
example, are stimulated, too. Due to their low chronaxie value, they are assumed to be
the elements primarily activated by DBS. Depending on distance and orientation with
respect to the stimulation contacts, DBS triggers action potentials in some axons but not
in others. These may travel either towards the synaptic terminals (orthodromic
activation) or towards the soma (antidromic activation). The observed effects of
orthodromic activation naturally depend on the types (excitatory or inhibitory) and the
number of synapses separating stimulation target and the site of measurement.
Given the wide range of possible effects and their dependence on the exact electrode
placement, it is not surprising that electrophysiological studies have not yet provided a
conclusive characterization of the mechanism underlying DBS. Interestingly, a recent
optpogenetic study proposed that out of the manifold of effects it is stimulation of
cortical STN afferents which causes clinical improvement (Gradinaru et al., 2009).
22 22Introduction
1.5 Oscillations in Parkinson’s disease
Implantation of DBS electrodes provides the unique opportunity to directly measure
neuronal activity from the human basal ganglia. Such recordings can either be
performed during surgery or in the interval between electrode and stimulator
implantation. If the surgeon uses microelectrodes to locate the target, extracellular
recordings from individual neurons can be obtained during surgery. In most studies,
however, the DBS electrodes are used for recordings. Due to their large contacts, DBS
electrodes record local field potentials (LFPs) rather than single cell activity. LFPs
represent the spatial average of electric fields over several hundred micrometers
(Katzner et al., 2009) and are believed to predominantly represent synaptic currents
rather than action potentials (reviewed in Buzsáki et al., 2012).
In both human PD patients and animal models of PD, LFP recordings revealed prominent
oscillatory activity in the basal ganglia. These discoveries led to a shift of attention away
from firing rates and towards rhythmic activity.
1.5.1 Alpha oscillations
Following treatment with MPTP, some monkey species develop strong alpha oscillations
in the basal ganglia (Bergman et al., 1994). The peak frequency of these oscillations is
usually twice the frequency of the MPTP-induced tremor. Thus, alpha oscillations are
believed to be related to tremor. In PD patients, similar tremor-related alpha oscillations
were observed in GPi (Hutchison et al., 1997) and STN (Levy et al., 2000). These
oscillations were coherent with tremor recordings from the muscle obtained by
electromyography (EMG), supporting their possible involvement in tremor generation.
Before this thesis was conducted, however, it was not known whether alpha oscillations
are directly related to the presence and/or severity of tremor.
In addition to their role in tremor, alpha oscillations in the basal ganglia are modulated
by voluntary movement. Recordings from the STN of PD patients demonstrated a
reduction of alpha power which starts 2 s prior to movement and continues during
movement execution (Oswal et al., 2012).
23 23Introduction
1.5.2 Beta oscillations
Recordings from the human STN and GPi revealed strong beta oscillations in PD
patients. Beta power was found to be reduced by both levodopa administration (Brown
et al., 2001) and by DBS (Eusebio et al., 2011). Moreover, the levodopa-induced
reduction of beta power was reported to correlate with the concurrent reduction of
akinesia and rigidity (Kühn et al., 2006). In addition, stimulation in the beta band was
found to slow movement. Slowing was achieved both by STN DBS (Chen et al., 2007)and
transcranial alternating current stimulation over motor cortex (Pogosyan et al., 2009).
Together, these findings led to the hypothesis that beta oscillations are pathologically
enhanced in PD. In addition, it was suggested that pathological hyper-synchrony causes
akinesia and rigidity (Brown, 2007). A potential mechanism was suggested based on
computational studies (Bar-Gad et al., 2003): enhanced synchrony might interfere with
de-correlation of cortical input in the basal ganglia and thereby impair information
compression (Hammond et al., 2007).
The idea of pathological hyper-synchrony in the beta band is popular in the field and has
inspired numerous experiments. However, it has also been criticized. While it is
generally agreed that STN beta power is a biomarker for akinesia, some studies cast
doubt on the causal role of beta oscillations. The reported effect sizes for symptom
worsening through stimulation at beta frequencies were small (Chen et al., 2007;
Pogosyan et al., 2009). Moreover, the effect could not be reproduced in a recent animal
experiment (Syed et al., 2012). Finally, akinesia was found to emerge before enhanced
beta oscillations appeared in non-human primates (Leblois et al., 2007).
1.5.3 Gamma and high frequency oscillations
Gamma oscillations were proposed to have a “prokinetic” function and to be the
functional counterpart of beta oscillations. In contrast to STN beta power, STN gamma
power increases following levodopa administration, both at rest and during movement
execution (Williams et al., 2002; Litvak et al., 2012). Moreover, beta and gamma
oscillations show antagonistic modulations during movement (Fig. 2). While beta power
is reduced prior to and during movement, STN gamma power and STN-M1 gamma
coherence increase at movement onset (Cassidy et al., 2002; Litvak et al., 2012).
24 24Introduction
Recently, a fourth frequency band attracted attention: high frequency oscillations (HFOs;
>200 Hz) were observed in the STN of medicated PD patients (Foffani et al., 2003). HFOs
were found to be phase-amplitude coupled to STN beta oscillations in a dopamine-
dependent fashion (López-Azcárate et al., 2010). Off medication, there was strong beta-
HFO coupling and little HFO amplitude modulation by movement. On medication, cross-
frequency coupling was reduced and HFO amplitude was strongly modulated during
movement. In consequence, it was proposed that beta oscillations impair physiological
HFO modulation through cross-frequency coupling. Another study confirmed the phase-
amplitude coupling between beta oscillations and HFOs and investigated the clinical
relevance of HFOs (Özkurt et al., 2011). The authors proposed a subdivision of the HFO
band into a fast and a slow HFO band and showed that the ratio between fast and slow
HFO power is reliably increased by levodopa administration.
Fig 2: Movement-induced power changes in the STN of Parkinson patients. Left: The figure illustrates the antagonistic relationship between beta and gamma oscillations in the STN. Beta power is suppressed relative to baseline immediately before and during a button press. Accordingly, beta oscillations are sometimes referred to as “antikinetic” oscillations. Following movement execution, beta power increases (beta
rebound) and then returns to baseline. In contrast, gamma power increases at movement onset and immediately returns to baseline. Thus, gamma oscillations are often considered “prokinetic”. Right: Like gamma oscillations, high frequency oscillations increase in power at movement onset. The increase is sustained for a longer period than for gamma oscillations. Colors indicate percent change relative to baseline (-8 to -5 s). Adapted from Litvak et al. (2012).
25 25Aims
2 Aims
The vast majority of studies on oscillations in PD dealt with synchrony within the STN.
Given the network structure of the motor system, however, it seems unlikely that
behaviorally relevant oscillatory activity is restricted to a single nucleus. In fact,
previous studies demonstrated that STN oscillations are coherent with oscillations in
other parts of the basal ganglia-cortical loop, such as the GPi (Brown et al., 2001).
Importantly, they were also found to be coherent with cortical oscillations (Williams et
al., 2002). However, the exact cortical areas coupled to the STN were not known.
Likewise, it was unclear whether and how STN-cortical coherence is modulated by
movement, levodopa administration and tremor.
In order to answer these open questions, the current thesis investigated coherence
between STN, cortex and muscle under various experimental conditions and in different
patient cohorts. The overall aim was to characterize synchronous oscillations in PD on
the network level. Specifically, the studies aimed at:
Study 1: Identifying the cortical areas coupled to the STN at rest. In particular, it was
investigated whether the spatial pattern of STN-cortical coherence is different
for different frequency bands.
Study 2: Investigating the effect of movement and levodopa administration on STN-
cortical coherence. Furthermore, modulations of STN-cortical coherence were
compared to modulations of cortico-muscular coherence in order to assess
possible dependencies between these two couplings.
Study 3: Describing the relationship between coherence and the manifestation of rest
tremor. Specifically, the study aimed at clarifying whether coherence at tremor
frequency increases when rest tremor emerges spontaneously.
26 26Paradigm
3 Paradigm
All three studies presented in this thesis made use of essentially the same experimental
setup and design. Three types of signals were recorded simultaneously: LFPs from the
STN, MEG and the EMG of forearm muscles. The LFP signal was recorded by two
bilaterally implanted DBS electrodes which were used for recording rather than
stimulation. Each DBS electrode had four contacts (0-3; 0 most ventral, 3 most dorsal).
The MEG signal was recorded by a 306-channel, whole-head MEG system. EMG was
recorded by surface electrodes over the extensor digitorum communis and flexor
digitorum superficialis muscles. Recordings were performed in the interval between
electrode and stimulator implantation. Importantly, deep brain electrodes were
connected to the amplifiers integrated in the MEG system by non-magnetic extension
leads so that the severe artifacts described in previous LFP-MEG studies were avoided
(Litvak et al., 2010).
The experiment consisted of two blocks. The first block was recorded after withdrawal
of dopaminergic medication (OFF). The second block was recorded after levodopa
administration (ON). Each block was preceded by a clinical rating of motor symptom
severity using the motor score of the Movement Disorder Society Unified Parkinson’s
Disease Rating Scale (MDS UPDRS III). Within each block, two recordings were
performed (Fig. 3). Both of these consisted of a 5 min rest period (REST) followed by one
of two motor tasks. The tasks were performed with the symptom-dominant hand. In the
first task (HOLD), subjects were asked to elevate their forearm to about 45° and to
spread their fingers. The elbow was leaning on a table in front of them. In the second
task (MOVE), subjects were instructed to open and close their fist repetitively. The
forearm was elevated as in the HOLD condition. Movements were self-paced and
performed with a frequency of approximately 1 Hz. Each motor task was interleaved by
pauses in order to avoid muscle fatigue. More specifically, epochs of task execution of 1
min duration and pauses of 1 min duration alternated for 9 minutes in total (Fig. 3).
The studies differed with respect to the epochs that were analyzed and with respect to
patient cohort. Study 1 investigated oscillatory coupling in the REST OFF condition and
included akinetic-rigid PD patients. Study 2 assessed oscillatory coupling during motor
task performance in the same patients (plus two additional subjects) in OFF and ON.
27 27
Study 1: Distinct oscillatory STN-cortical loops revealed by simultaneous MEG and local
field potential recordings in patients with Parkinson's disease
Study 3 focused on the same epoch as study 1 but included tremor-dominant PD
patients.
Fig. 3: Schematic time line for one block of simultaneous LFP-MEG recordings. A block (either OFF or ON) consisted of two recordings which were performed in
succession, but are depicted as parallel lines for illustration purposes. Each recording
contained a 5 min rest period (REST) followed by a motor task. The task was either to
elevate the forearm to 45° and to keep this position (HOLD) or to open and close the fist
repeatedly with approximately 1 Hz (MOVE). Epochs of task execution were interleaved
by 1 min pauses in order to avoid muscle fatigue. In studies 1 and 3, the REST OFF period was analyzed. Study 2 focused on the periods of
motor task performance in medication OFF and ON.
4 Study 1: Distinct oscillatory STN-cortical loops revealed by
simultaneous MEG and local field potential recordings in patients
with Parkinson's disease
Study 1 (Appendix 1) investigated coherence between STN and cortex. STN-cortical
coherence is an especially interesting parameter as the cortex was shown to drive STN
oscillations in the beta band (Williams et al., 2002; Lalo et al., 2008; Litvak et al., 2011).
Therefore, it is often speculated that pathologically enhanced beta activity originates in
the cortex. When study 1 was performed, it was known that there is significant
coherence between STN and cortex (Williams et al., 2002; Fogelson et al., 2006).
28 28
Study 1: Distinct oscillatory STN-cortical loops revealed by simultaneous MEG and local
field potential recordings in patients with Parkinson's disease
However, it was unclear exactly which cortical areas are involved in this coupling and
whether different cortical areas are coupled to the STN at different frequencies. Study 1
aimed at answering these open questions.
An earlier EEG study had already found first indications for a frequency-specific spatial
distribution of STN-cortical coherence (Fogelson et al., 2006). However, this study was
restricted to the sensor level because post-surgical dressing precluded measurements
with more than a few EEG electrodes. In consequence, source localization was not
feasible. Study 1 mapped STN-cortical coherence on the source level using MEG which
allows for high density measurements even in the presence of surgical dressing.
4.1 Methods
Nine PD patients of the akinetic-rigid subtype participated in the study. Patients with
tremor were not included since tremor is associated with strong alpha oscillations
which potentially mask resting state alpha oscillations (Timmermann et al., 2003). One
patient was excluded due to extensive head movement artifacts. STN LFPs, MEG and
forearm EMG were recorded simultaneously. The study investigated the REST OFF
condition (see Fig. 3). Four frequency bands were investigated: alpha (7-12 Hz), low
beta (13-20 Hz), high beta (21-35 Hz) and gamma (70-90 Hz). For each frequency band,
patient and electrode contact, the frequency with maximum coherence between STN
LFPs and the MEG signal was determined automatically on the sensor level. DICS (Gross
et al., 2001) was then applied for this frequency to localize coherence.
4.2 Results
4.2.1 Distribution of STN-cortical coherence across brain areas
All subjects showed significant coherence peaks in the alpha and beta band. Except for
one case, gamma band coherence was not observed. Alpha coherence localized to a
number of brain areas ipsilateral to the STN (Fig. 6 of Appendix 1). Although there was
no significant overlap of alpha sources across subjects, a cluster of sources was observed
in temporal cortex, in particular in superior temporal gyrus (STG). In contrast, beta
sources consistently localized to medial sensorimotor and premotor areas ipsilateral to
the STN.
29 29
Study 1: Distinct oscillatory STN-cortical loops revealed by simultaneous MEG and local
field potential recordings in patients with Parkinson's disease
4.2.2 Distribution of STN-STG and STN-M1 coherence across electrode
contacts
In a subsequent analysis step, M1 and STG were defined as regions of interest (ROIs) and
the distribution of STN-ROI coherence across STN electrode contacts was evaluated.
STN-STG coherence peaked in the alpha band. The distribution of alpha peaks across
contacts was not significantly different from a uniform distribution (Fig. 7 of Appendix
1), indicating that STG oscillations are coherent with a subcortical alpha source with
large spatial extent. In contrast, STN-M1 coherence peaked in the beta band and beta
peaks were usually observed for only one or two contacts, i.e. STN-M1 beta coherence
was focal within the area recorded by the electrode.
4.3 Discussion
Study 1 showed that the spatial distribution of cerebral coherence with STN LFPs is
frequency-dependent. STN alpha oscillations were coherent with oscillatory activity in
temporal areas whereas STN beta oscillations were coherent with oscillations in
sensorimotor and premotor cortex. Thus, different spectral components of the same
subcortical signal coupled to different cortical areas. Notably, the frequency-dependent
spatial pattern found in study 1 was confirmed by an independent study which also
investigated STN-cortical coherence in PD patients (Litvak et al., 2011).
The function of the separation of couplings in the frequency domain remains unclear.
One possibility is that inputs from M1 and inputs from STG are distinguished in the STN
based on the frequency of the incoming signal. This strategy is known as multiplexing in
communications engineering and usually serves to avoid interference between signals
which are transmitted via the same physical channel (Weinstein and Ebert, 1971).
The observation of strong beta coherence between STN and motor cortex is plausible
with regard to anatomy. The STN receives polysynaptic input from ipsilateral motor
cortex via the indirect pathway and projects back to motor cortex via the thalamus (see
chapter 2.1). Moreover, it receives monosynaptic cortical input via the hyperdirect
pathway. A recent study aimed at imaging the hyperdirect pathway in humans by
diffusion tensor imaging (Whitmer et al., 2012). The authors reported that the pathway
originates in medial motor cortex, i.e. at a location comparable to the site of maximum
beta coherence observed in the present study. They placed several strip electrodes onto
30 30
Study 2: Differential modulation of STN-cortical and cortico-muscular coherence by
movement and levodopa in Parkinson's disease
the cortical surface and found that beta coherence with STN was elevated in those
contacts which covered the previously identified origin of the alleged hyperdirect
pathway. Thus, intracranial recordings confirmed the results obtained in this study and
suggest that the STN-cortical beta coherence reported here reflects direct motor cortical
input to the STN. This interpretation tallies with the focal spatial distribution of STN-M1
beta coherence across electrode contacts. Both motor cortex and STN are
somatotopically organized and any given cortical motor area projects precisely to its
counterpart STN region (Nambu et al., 1996). Thus, any area in motor cortex is expected
to be coupled to a limited portion of the STN rather than the whole nucleus.
The anatomical basis and the possible function are less clear for STN-STG alpha band
coupling. A recent MEG-LFP study confirmed its existence and found it to be modulated
by movement and dopaminergic medication (Oswal et al., 2012). Since STN-STG alpha
coherence was not affected by the specific type of motor task, it was hypothesized that
the coupling reflects a default functional interaction which needs to be transiently
terminated before and during any kind of movement.
4.4 Conclusions
The spatial distribution of STN-cortical coherence in PD patients is frequency-
dependent. There is a beta and an alpha pattern. The beta pattern represents functional
connectivity between STN and cortical motor areas which might be facilitated by the
hyperdirect pathway. The alpha pattern represents functional connectivity between STN
and temporal areas. Its function and anatomical basis need further investigation.
5 Study 2: Differential modulation of STN-cortical and cortico-
muscular coherence by movement and levodopa in Parkinson's
disease
The aim of study 2 (Appendix 2) was to explore the effects of movement and
dopaminergic medication on the oscillatory network identified in study 1. The
experiment was designed to test hypotheses put forward by Peter Brown and colleagues
who suggested that pathological hyper-synchrony in the beta band causes the slowing of
movement in PD (Brown, 2007). Furthermore, it built on interpretations by Engel and
Fries who consider beta oscillations a neural correlate of maintaining the status quo
31 31
Study 2: Differential modulation of STN-cortical and cortico-muscular coherence by
movement and levodopa in Parkinson's disease
(Engel and Fries, 2010). Based on these ideas, it was hypothesized that administration of
levodopa will reduce beta coherence along with restoring motor capabilities. Moreover,
movement was expected to decrease beta coherence, as movement implies a change in
motor state.
The relationship between STN-cortical and cortico-muscular beta coherence was of
particular interest. Given the generality of the hypotheses outlined above, one would
expect beta band coupling to show the same responses everywhere in the motor system.
Thus, experimental manipulations such as motor task execution or levodopa
administration should elicit the same changes in STN-cortical and cortico-muscular beta
coherence. However, it was reported previously that medication affects these two
couplings differentially. Administration of levodopa was found to decrease STN-cortical
beta coherence (Williams et al., 2002; Sharott et al., 2005) but to increase cortico-
muscular beta coherence (Salenius et al., 2002). Thus, there are indications that STN-
cortical and cortico-muscular beta coherence are independent to some degree. However,
the levodopa-induced decrease in STN-cortical beta coherence is not a consistent
finding. It was observed in some studies (Williams et al., 2002; Sharott et al., 2005) but
not in others (Lalo et al., 2008; Litvak et al., 2011). In summary, the available data on the
relationship between STN-cortical and cortico-muscular coherence is inconclusive. One
reason for the divergent results could be that they were obtained in different patient
cohorts. In order to perform a direct comparison, STN-cortical and cortico-muscular
coherence were measured in the same patients in study 2.
5.1 Methods
10 PD patients of the akinetic-rigid subtype participated in the study. STN LFPs, MEG
and forearm EMG were recorded simultaneously. Subjects performed two motor tasks in
succession (see section 3): continuous elevation of the forearm (HOLD) and repetitive
opening and closing of the fist (MOVE). Both tasks were performed once after
withdrawal of anti-parkinsonian medication (OFF) and once after administration of
levodopa (ON).
Coherence with STN LFPs and EMG was computed for two cortical regions of interest
(ROIs). These were chosen as the coherence group maxima identified in study 1, i.e. M1
and STG. Coherence was analyzed in three frequency bands: alpha (8-12 Hz), beta (13-
32 32
Study 2: Differential modulation of STN-cortical and cortico-muscular coherence by
movement and levodopa in Parkinson's disease
35 Hz), and gamma (60-90 Hz). Effects of movement (factor levels: HOLD and MOVE)
and medication (factor levels: ON and OFF) on coherence were tested by repeated
measures analysis of variance. To investigate a potential relationship between
coherence and PD symptoms, coherence was correlated with UPDRS akinesia and
rigidity sub-scores.
5.2 Results
5.2.1 Effects of movement and medication
Coherence with M1 but not with STG was responsive to movement and medication.
Interestingly, STN-M1 and M1-muscular coherence were modulated differentially (Fig. 1
of Appendix 2). M1-muscular coherence was decreased by repetitive movement
compared to static contraction in the alpha and beta band. However, it was not
modulated by levodopa administration. In contrast, STN-M1 beta coherence was
decreased by levodopa administration but not significantly altered by movement.
5.2.2 Correlation with clinical parameters
Surprisingly, the dopamine-induced decrease in STN-M1 beta coherence was not
correlated with the decrease in akinesia and rigidity scores. However, absolute STN-M1
beta coherence and akinesia/rigidity UPDRS scores were negatively correlated in the
OFF state, i.e. the subjects with the strongest coherence had the least akinesia (Fig. 5 of
Appendix 2). The correlation was specific to the beta frequency band and the OFF state,
but unspecific with regard to motor task. A qualitatively similar result was obtained for
M1-muscular beta coherence, suggesting inter-dependence between STN-M1 and M1-
muscular coupling. Indeed, STN-cortical and cortico-muscular beta coherence were
positively correlated in all experimental conditions, except for HOLD ON (Fig. S5 of
Appendix 2).
5.3 Discussion
Study 2 revealed that STN-cortical and cortico-muscular coherence are differentially
modulated by movement and medication, suggesting that they represent two partly
independent functional loops. Please note, however, they are not entirely independent.
33 33
Study 2: Differential modulation of STN-cortical and cortico-muscular coherence by
movement and levodopa in Parkinson's disease
In most of the experimental conditions, they were positively correlated. The current
results rather suggest that the two couplings respond differentially to levodopa but are
closely related otherwise.
Previous studies investigating the effect of movement on M1-muscular (Kilner et al.,
2000) and STN-M1 coherence (Litvak et al., 2012) found similar modulations. Both
couplings showed a gamma increase at movement onset and a beta rebound following
movement termination. Likewise, the current study found mean STN-cortical beta
coherence to be reduced in the MOVE compared to the HOLD condition, but the
reduction was less pronounced than for cortico-muscular coherence and not significant.
Thus, it seems reasonable to assume that movement modulates STN-cortical and
cortico-muscular coherence in a similar way. However, the modulation of cortico-
muscular beta coherence appears to be stronger.
In contrast to movement, levodopa administration clearly had different effects on
cortico-muscular and STN-cortical coupling. It reduced STN-cortical beta coherence but
did not affect cortico-muscular beta coherence. The levodopa-induced reduction of STN-
cortical beta coherence is in line with the concept of pathological hyper-synchrony in the
motor system of PD patients (Brown, 2007). However, the negative correlation with
akinesia and rigidity scores speaks against this hypothesis. Rather than being associated
with akinesia, strong beta coherence appears to reflect relatively good mobility in the
dopamine-depleted state.
There are two possible explanations for the obtained results: STN-cortical beta
coherence might reflect a compensatory process which promotes movement in the OFF
state but becomes oblivious when normal basal ganglia functionality is restored by
levodopa administration. This interpretation would explain why the negative correlation
with akinesia and rigidity was not observed in medication ON.
Alternatively, it is conceivable that STN-cortical beta coherence, as opposed to STN beta
power, is required for normal motor function and is abnormally reduced in PD patients.
The additional reduction induced by levodopa administration might be a side-effect of
medication. It could result from the strong reduction of STN beta power and the
resulting drop in signal-to-noise ratio.
34 34
Study 2: Differential modulation of STN-cortical and cortico-muscular coherence by
movement and levodopa in Parkinson's disease
A recent measurement performed in our laboratory supports the notion of different
roles of STN beta power and STN-cortical beta coherence in PD pathophysiology
(unpublished data). A patient with obsessive compulsive disorder (OCD) was implanted
with bilateral electrodes for STN DBS and measured with the same experimental setup
as used in this thesis. The experiment provided the unique opportunity to compare PD
patients to a control subject free of motor impairments. Interestingly, this patient’s STN-
cortical beta coherence was very similar to coherence in PD patients, both with respect
to the spatial distribution and with respect to coupling strength. In contrast, STN beta
power was markedly lower than in any of the PD patients. In summary, these results
suggest that the spatial pattern of STN-cortical beta coherence observed in PD patients
is of physiological nature. Moreover, PD patients seem to exhibit strong STN beta power
but normal to weak STN-cortical beta coherence, indicating that beta power and beta
coherence are functionally different. These conclusions should be treated with caution,
however, since the variability across individuals without movement disorder remains
unknown.
5.4 Conclusions
Overall, study 2 suggests that oscillatory coupling in the beta band does not respond in a
homogeneous fashion everywhere in the motor system. STN-cortical beta coherence was
reduced by levodopa administration while cortico-muscular beta coherence was not
affected. Furthermore, the negative correlation between beta coherence and akinesia
demonstrates that beta band synchrony within the motor system does not impair
movement execution in general. In contrast to local beta synchrony within the STN,
inter-regional beta synchrony between STN and M1 is associated with comparably good
motor performance in the dopamine-depleted state.
35 35
Study 3: A direct relationship between oscillatory STN-cortex coupling and rest tremor in
Parkinson’s disease
6 Study 3: A direct relationship between oscillatory STN-cortex
coupling and rest tremor in Parkinson’s disease
Besides akinesia and rigidity, tremor is one of the most eminent and frequent symptoms
in PD. Classical parkinsonian tremor has a frequency of 3-7 Hz, occurs at rest and is
attenuated at movement onset (reviewed in Deuschl et al., 2000). When patients settle
to a new static position, the tremor typically reappears within a few seconds. At rest,
tremor is usually not continuously present but intermitted by spontaneous pauses.
In the 1990s, it was debated whether tremor is caused by central or peripheral
mechanisms such as spinal reflex arcs or mechanical resonances. During the last two
decades, evidence for central oscillatory mechanisms has accumulated (reviewed in
McAuley and Marsden, 2000; Schnitzler and Gross, 2005; Schnitzler et al., 2006).
Electrophysiological studies revealed neuronal oscillations in STN, GPi and thalamus
which were coherent with EMG recordings at tremor frequency and its first upper
harmonic (e.g. Hurtado et al., 2005; Reck et al., 2009). A more extended tremor network
including subcortical and cortical areas was identified by Timmermann et al. (2003)
who localized cerebro-muscular coherence using MEG. The same network was later
found to underlie voluntary tremor in healthy subjects (Pollok et al., 2004). Thus, there
is evidence for pathological synchronization between several cortical and subcortical
areas and tremulous muscles in parkinsonian rest tremor.
While power and coherence peaks at tremor frequency are good indications for central
mechanisms in tremor generation, matching frequency alone does not imply correlation
between neuronal oscillations and tremor manifestation − let alone a causal
relationship. Theoretically, neuronal oscillations and tremor might be independent. In
fact, rhythmic bursting at 5 Hz has been observed in the STN of tremor-free patients
(Magariños-Ascone et al., 2000). Study 3 (Appendix 3) aimed at providing evidence for a
direct relationship between coherence and tremor manifestation. In other words, it
investigated whether coherence at tremor frequency is indeed a neural correlate of rest
tremor. In contrast to previous studies, it compared tremor epochs to tremor-free
epochs within subjects to clarify i) whether coherence increases when tremor emerges
and ii) whether coherence is correlated with tremor amplitude.
36 36
Study 3: A direct relationship between oscillatory STN-cortex coupling and rest tremor in
Parkinson’s disease
6.1 Methods
11 tremor-dominant PD patients participated in the study. STN LFPs, MEG and forearm
EMG were recorded simultaneously. The study compared epochs with spontaneous rest
tremor in the upper limb to tremor-free epochs in the REST OFF condition (Fig. 3). STN
power, MEG power, STN-cortical coherence, cortico-muscular coherence and STN-
muscular coherence were estimated on the sensor level. MEG channels of interest were
selected a priori such that they covered motor and premotor areas contralateral to the
tremulous limb. This selection was refined individually based on maximal sensor power
at tremor frequency (conditions pooled). Similarly, LFP and EMG channel selection was
based on maximal coherence at tremor frequency with the MEG sensors of interest.
Following sensor level analysis, coherence changes induced by tremor were localized
using DICS (Gross et al., 2001). Cortico-cortical source level coupling was investigated by
analysis of coherence and the imaginary part of coherency, a coupling measure
insensitive to volume conduction (Nolte et al., 2004).
6.2 Results
6.2.1 Sensor level
A time-resolved analysis of spectral power revealed that STN oscillations at individual
tremor frequency increase in amplitude when tremor emerges (Fig. 2 of Appendix 3). In
addition, cortical power decreased in the beta band following tremor onset. STN-
cortical, cortico-muscular and STN-muscular coherence were found to be higher during
tremor than during tremor-free epochs specifically at tremor frequency (Fig. 3 of
Appendix 3). The tremor-induced change in STN-cortical coupling was positively
correlated with the change in EMG power.
6.2.2 Source level
Analysis of source level coherence at tremor frequency revealed significant changes in
cortico-muscular but not in STN-cortical coherence. Tremor-induced increases were
observed in M1, premotor cortex (PMC) and posterior parietal cortex (PPC)
contralateral to the tremulous limb (Fig. 4 of Appendix 3). These cortical areas were
found to be themselves coupled at tremor frequency and double tremor frequency. Like
37 37
Study 3: A direct relationship between oscillatory STN-cortex coupling and rest tremor in
Parkinson’s disease
cortico-muscular coupling, cortico-cortical coupling increased during tremor.
Importantly, analysis of the imaginary part of coherency showed that cortico-cortical
coupling was not a trivial consequence of volume conduction.
6.3 Discussion
Study 3 demonstrated that oscillatory coupling within the sensorimotor system
increases during tremor. Moreover, it reproduced core elements of the previously
described tremor network. Like study 3, earlier studies reported tremor-related
coherence in M1, PMC and PPC (Timmermann et al., 2003; Pollok et al., 2004;
Muthuraman et al., 2012). Compared to these works, study 3 made use of a more
stringent methodology. The analysis included statistical tests at the group level and
controlled for multiple comparisons. Furthermore, it made use of a coupling measure
which is insensitive to trivial couplings resulting from volume conduction. Thus, it
provided a reliable description of the tremor network.
Naturally, increased reliability comes at the cost of decreased sensitivity. Study 3 did not
detect tremor-induced coherence changes in thalamus or cerebellum although both are
likely involved in tremor generation (Helmich et al., 2011; Mure et al., 2011). Moreover,
it did not detect an increase in STN-cortical coherence on the source level even though it
was observable on the sensor level. Most likely, recording times were too short to
localize this effect.
Study 3 provided important insights into the role of the STN in tremor. It confirmed a
single case report stating that beta power decreases as tremor becomes manifest (Wang
et al., 2005). In addition, it showed that STN power at individual tremor frequency
increases following tremor onset. Thus, study 3 provided an electrophysiological
biomarker of tremor onset which could be used by closed-loop DBS systems to trigger
stimulation (see section 8). Further, study 3 revealed that the coupling between STN and
other brain areas increases during tremor. Importantly, the increase was proportional to
the increase in EMG power, suggesting that the STN generates or receives oscillatory
signals which directly reflect tremor amplitude.
38 38General discussion
6.4 Conclusions
Study 3 established that STN, primary motor cortex, premotor cortex and posterior
parietal cortex are part of the tremor network. Furthermore, it demonstrated that
coherence within this network increases during tremor. In conclusion, the results
suggest that oscillatory coupling at tremor frequency is indeed a neural correlate of
parkinsonian rest tremor.
7 General discussion
This thesis aimed at investigating synchronous oscillations in PD on the network level by
scanning the brain for coherence with STN LFPs and muscle activity, and by
characterizing changes in coherence due to movement, dopaminergic medication and
tremor. Moreover, it explored the relationship between neuronal oscillations and PD
symptoms.
The overall impact was twofold: First, the studies substantially advanced the
establishment of simultaneous MEG-LFP recordings as a tool for investigating functional
connectivity in PD (reviewed in Schnitzler and Hirschmann, 2012; Appendix 4). The
technique was introduced not more than three years ago (Litvak et al., 2010). Due to the
requirements with regard to hardware and patient access, it is available in just a few
centers worldwide. Thus, the studies presented here can be understood as a proof of
principle, demonstrating the usefulness and feasibility of the approach.
Second, the thesis contributed significantly to the characterization of oscillatory
coupling in PD. In addition to revealing the frequency-dependent spatial distribution of
coherence between STN and cortex (Study 1), it confirmed some common ideas about
the role of neuronal oscillations in PD and motor control in general. In line with the
hypothesis that beta oscillations signal maintenance of the status quo (Engel and Fries,
2010), beta coherence between motor cortex and muscle was found to be suppressed
during repetitive movement compared to static contraction (study 2). Moreover, tremor
manifestation was found to be associated with an increase in coherence at tremor
frequency in a distributed sensorimotor network (study 3), supporting the hypothesis
that pathologically enhanced neuronal synchronization underlies parkinsonian rest
tremor (Schnitzler et al., 2006). However, not all of the results are in line with the idea of
39 39Outlook
pathological hyper-synchrony. STN-cortical beta coherence was negatively correlated
with akinesia and rigidity UPDRS sub-scores (study 2), suggesting that it counteracts
rather than promotes the slowing of movement. Currently, there are no other
publications to confirm or disconfirm this observation. Thus, future studies are needed
to clarify whether STN-cortical beta coherence reflects a pathological or a physiological
mechanism. Confirmation of the physiological or compensatory function suggested by
study 2 would imply that STN beta power and STN-cortical beta coherence reflect
fundamentally different processes.
In summary, this thesis made valuable contributions to research on PD pathophysiology.
It showed that coherent oscillations are modulated during movement, respond to
medication and correlate with clinical symptoms. These findings substantiate the role of
oscillatory coupling in PD.
8 Outlook
The above discussion about the function of STN-cortical beta coherence highlights a
fundamental challenge to clinical electrophysiology: biomarkers need to be separated
from physiological mechanisms and causes need to be separated from consequences of
the disease. Distinguishing biomarkers from physiological mechanisms is especially
challenging in invasive patient studies as healthy controls are lacking. At best,
recordings from PD patients can be compared to recordings from patients with
unrelated diseases. In most cases, however, inference on pathological relevance is based
on treatment ON vs. treatment OFF contrasts or experiments in animal models. Both
approaches bear obvious caveats: pharmacological as well as surgical treatment does
not only have clinical, but also electrophysiological side-effects. In consequence, changes
observed in the treatment ON condition might be unrelated to the treatment-induced
improvement of symptoms. With regard to animal models, interpretational difficulties
arise from symptomatic, electrophysiological and anatomical differences between PD
patients and MPTP-treated monkeys and 6OHDA-treated rodents, respectively (Betarbet
et al., 2002).
In the future, PD research will need to identify reliable, electrophysiological biomarkers
of PD in humans and clarify whether these are causal or correlative. In particular, it is
important to know if hyper-synchrony may indeed cause motor impairments. Apart
40 40Outlook
from entrainment of neuronal oscillations by periodic brain stimulation, such as DBS at
beta frequencies, one approach could be to monitor oscillations for longer periods of
time. Longitudinal studies can answer an important question: What comes first,
symptoms or synchronization?
So far, longitudinal electrophysiological studies are rare and restricted to animal models
of PD. Interestingly, the available data suggest that beta synchrony and motor symptoms
do not develop in parallel, indicating that beta oscillations might not cause motor
symptoms (Leblois et al., 2007; Degos et al., 2009; Dejean et al., 2012). Future studies
need to elaborate on the temporal evolution of electrophysiological hallmarks and relate
it to disease progression. In patients, this will be possible in the near future. The new
generation of DBS devices will be able to record data from deep brain electrodes and
transmit them via radio communication (Santa et al., 2008). Thus, physical access to the
electrodes will no longer be needed for deep brain recordings and the long-term
dynamics of oscillations can be investigated in humans.
Apart from radio communication, the new generation of DBS systems will most likely be
able to register brain activity and to flexibly react to changes in electrophysiological
parameters. A recent animal study provided first evidence that closed-loop stimulation
triggered by endogenous brain activity provides better symptom alleviation than
conventional DBS (Rosin et al., 2011). Since closed-loop systems are not continuously
active, they are expected to reduce the occurrence of side-effects and to provide more
efficient battery usage. One of the challenges in the development of closed-loop systems
is the identification of appropriate control parameters. This thesis suggests that
neuronal oscillations might be a good candidate. More specifically, improved tremor
alleviation might be achieved by applying DBS whenever STN power increases at double
the tremor frequency and decreases in the beta band. Ideally, stimulation would be
tailored to obtain destructive interference with the endogenous tremor rhythm (Brittain
et al., 2013). Moreover, the current thesis indicates that resting state STN beta power,
but not STN-cortical beta coherence, is a promising candidate parameter which could be
used for triggering high frequency DBS in patients suffering from akinesia.
41 41References
9 References
Albin, R.L., Young, A.B., and Penney, J.B. (1989). The functional anatomy of basal ganglia disorders. Trends in Neurosciences 12, 366–375.
Alexander, G.E., DeLong, M.R., and Strick, P.L. (1986). Parallel Organization of Functionally Segregated Circuits Linking Basal Ganglia and Cortex. Annual Review of Neuroscience 9, 357–381.
Ashby, F.G., Turner, B.O., and Horvitz, J.C. (2010). Cortical and basal ganglia contributions to habit learning and automaticity. Trends in Cognitive Sciences 14, 208–
215.
Bar-Gad, I., Morris, G., and Bergman, H. (2003). Information processing, dimensionality reduction and reinforcement learning in the basal ganglia. Progress in Neurobiology 71, 439–473.
Berger, P.D.H. (1929). Über das Elektrenkephalogramm des Menschen. Archiv f. Psychiatrie 87, 527–570.
Bergman, H., Wichmann, T., and DeLong, M.R. (1990). Reversal of experimental parkinsonism by lesions of the subthalamic nucleus. Science 249, 1436–1438.
Bergman, H., Wichmann, T., Karmon, B., and DeLong, M.R. (1994). The primate subthalamic nucleus. II. Neuronal activity in the MPTP model of parkinsonism. J Neurophysiol 72, 507–520.
Betarbet, R., Sherer, T.B., and Greenamyre, J.T. (2002). Animal models of Parkinson’s
disease. BioEssays 24, 308–318.
Beurrier, C., Bioulac, B., Audin, J., and Hammond, C. (2001). High-Frequency Stimulation Produces a Transient Blockade of Voltage-Gated Currents in Subthalamic Neurons. J Neurophysiol 85, 1351–1356.
Braak, H., Tredici, K.D., Rüb, U., de Vos, R.A.., Jansen Steur, E.N.., and Braak, E. (2003). Staging of brain pathology related to sporadic Parkinson’s disease. Neurobiology of
Aging 24, 197–211.
Brittain, J., Probert-Smith, P., Aziz, T., and Brown, P. (2013). Tremor suppression by rhythmic transcranial current stimulation. Curr. Biol. 23, 436–440.
Brown, P. (2007). Abnormal oscillatory synchronisation in the motor system leads to impaired movement. Curr. Opin. Neurobiol. 17, 656–664.
Brown, P., Oliviero, A., Mazzone, P., Insola, A., Tonali, P., and Di Lazzaro, V. (2001). Dopamine dependency of oscillations between subthalamic nucleus and pallidum in Parkinson’s disease. J. Neurosci. 21, 1033–1038.
Buzsaki, G. (2006). Rhythms of the Brain (Oxford University Press).
Buzsáki, G., Anastassiou, C.A., and Koch, C. (2012). The origin of extracellular fields and currents — EEG, ECoG, LFP and spikes. Nature Reviews Neuroscience 13, 407–420.
42 42References
Cassidy, M., Mazzone, P., Oliviero, A., Insola, A., Tonali, P., Di Lazzaro, V., and Brown, P. (2002). Movement-related changes in synchronization in the human basal ganglia. Brain 125, 1235–1246.
Chen, C.C., Litvak, V., Gilbertson, T., Kühn, A., Lu, C.S., Lee, S.T., Tsai, C.H., Tisch, S., Limousin, P., Hariz, M., et al. (2007). Excessive synchronization of basal ganglia neurons at 20 Hz slows movement in Parkinson’s disease. Experimental Neurology 205, 214–221.
Degos, B., Deniau, J.-M., Chavez, M., and Maurice, N. (2009). Chronic but not Acute Dopaminergic Transmission Interruption Promotes a Progressive Increase in Cortical Beta Frequency Synchronization: Relationships to Vigilance State and Akinesia. Cereb. Cortex 19, 1616–1630.
Dejean, C., Nadjar, A., Le Moine, C., Bioulac, B., Gross, C.E., and Boraud, T. (2012). Evolution of the dynamic properties of the cortex-basal ganglia network after dopaminergic depletion in rats. Neurobiol. Dis. 46, 402–413.
DeLong, M.R. (1990). Primate models of movement disorders of basal ganglia origin. Trends in Neurosciences 13, 281–285.
Desmurget, M., and Turner, R.S. (2008). Testing Basal Ganglia Motor Functions Through Reversible Inactivations in the Posterior Internal Globus Pallidus. J Neurophysiol 99, 1057–1076.
Desmurget, M., and Turner, R.S. (2010). Motor Sequences and the Basal Ganglia: Kinematics, Not Habits. J. Neurosci. 30, 7685–7690.
Deuschl, G., Raethjen, J., Baron, R., Lindemann, M., Wilms, H., and Krack, P. (2000). The pathophysiology of parkinsonian tremor: a review. J. Neurol. 247 Suppl 5, V33–48.
Engel, A.K., and Fries, P. (2010). Beta-band oscillations--signalling the status quo? Curr. Opin. Neurobiol. 20, 156–165.
Eusebio, A., Thevathasan, W., Doyle Gaynor, L., Pogosyan, A., Bye, E., Foltynie, T., Zrinzo, L., Ashkan, K., Aziz, T., and Brown, P. (2011). Deep brain stimulation can suppress pathological synchronisation in parkinsonian patients. J. Neurol. Neurosurg. Psychiatr. 82, 569–573.
Foffani, G., Priori, A., Egidi, M., Rampini, P., Tamma, F., Caputo, E., Moxon, K.A., Cerutti, S., and Barbieri, S. (2003). 300-Hz subthalamic oscillations in Parkinson’s disease. Brain
126, 2153–2163.
Fogelson, N., Williams, D., Tijssen, M., van Bruggen, G., Speelman, H., and Brown, P. (2006). Different functional loops between cerebral cortex and the subthalmic area in Parkinson’s disease. Cereb. Cortex 16, 64–75.
Fries, P. (2005). A mechanism for cognitive dynamics: neuronal communication through neuronal coherence. Trends in Cognitive Sciences 9, 474–480.
Gerfen, C., Engber, T., Mahan, L., Susel, Z., Chase, T., Monsma, F., and Sibley, D. (1990). D1 and D2 dopamine receptor-regulated gene expression of striatonigral and striatopallidal neurons. Science 250, 1429–1432.
43 43References
Gerfen, C.R., and Bolam, J.P. (2010). The Neuroanatomical Organization of the Basal Ganglia. In Handbook of Basal Ganglia Structure and Function, H. Steiner, and K.Y. Tseng, eds. (Academic Press), pp. 3–28.
Gradinaru, V., Mogri, M., Thompson, K.R., Henderson, J.M., and Deisseroth, K. (2009). Optical Deconstruction of Parkinsonian Neural Circuitry. Science 324, 354–359.
Gross, J., Kujala, J., Hamalainen, M., Timmermann, L., Schnitzler, A., and Salmelin, R. (2001). Dynamic imaging of coherent sources: Studying neural interactions in the human brain. Proc. Natl. Acad. Sci. U.S.A 98, 694–699.
Hämäläinen, M., Hari, R., Ilmoniemi, R.J., Knuutila, J., and Lounasmaa, O.V. (1993). Magnetoencephalography—theory, instrumentation, and applications to noninvasive studies of the working human brain. Rev. Mod. Phys. 65, 413–497.
Hammond, C., Bergman, H., and Brown, P. (2007). Pathological synchronization in Parkinson’s disease: networks, models and treatments. Trends Neurosci. 30, 357–364.
Helmich, R.C., Hallett, M., Deuschl, G., Toni, I., and Bloem, B.R. (2012). Cerebral causes and consequences of parkinsonian resting tremor: a tale of two circuits? Brain 135, 3206–3226.
Helmich, R.C., Janssen, M.J.R., Oyen, W.J.G., Bloem, B.R., and Toni, I. (2011). Pallidal dysfunction drives a cerebellothalamic circuit into Parkinson tremor. Annals of Neurology 69, 269–281.
Hillebrand, A., Singh, K.D., Holliday, I.E., Furlong, P.L., and Barnes, G.R. (2005). A new approach to neuroimaging with magnetoencephalography. Hum Brain Mapp 25, 199–
211.
Hutchison, W.D., Lozano, A.M., Tasker, R.R., Lang, A.E., and Dostrovsky, J.O. (1997). Identification and characterization of neurons with tremor-frequency activity in human globus pallidus. Exp Brain Res 113, 557–563.
Jenkins, I.H., Fernandez, W., Playford, E.D., Lees, A.J., Frackowiak, R.S.J., Passingham, R.E., and Brooks, D.J. (1992). Impaired activation of the supplementary motor area in Parkinson’s disease is reversed when akinesia is treated with apomorphine. Annals of
Neurology 32, 749–757.
Katzner, S., Nauhaus, I., Benucci, A., Bonin, V., Ringach, D.L., and Carandini, M. (2009). Local Origin of Field Potentials in Visual Cortex. Neuron 61, 35–41.
Kilner, J.M., Baker, S.N., Salenius, S., Hari, R., and Lemon, R.N. (2000). Human cortical muscle coherence is directly related to specific motor parameters. J. Neurosci. 20, 8838–
8845.
Kincaid, A.E., Zheng, T., and Wilson, C.J. (1998). Connectivity and Convergence of Single Corticostriatal Axons. J. Neurosci. 18, 4722–4731.
Kravitz, A.V., Freeze, B.S., Parker, P.R.L., Kay, K., Thwin, M.T., Deisseroth, K., and Kreitzer, A.C. (2010). Regulation of parkinsonian motor behaviours by optogenetic control of basal ganglia circuitry. Nature 466, 622–626.
44 44References
Kringelbach, M.L., Jenkinson, N., Owen, S.L.F., and Aziz, T.Z. (2007). Translational principles of deep brain stimulation. Nature Reviews Neuroscience 8, 623–635.
Kühn, A.A., Kupsch, A., Schneider, G.-H., and Brown, P. (2006). Reduction in subthalamic 8-35 Hz oscillatory activity correlates with clinical improvement in Parkinson’s disease.
Eur. J. Neurosci. 23, 1956–1960.
Laitinen, L.V., Bergenheim, A.T., and Hariz, M.I. (1992). Leksell’s posteroventral
pallidotomy in the treatment of Parkinson’s disease. J. Neurosurg. 76, 53–61.
Lalo, E., Thobois, S., Sharott, A., Polo, G., Mertens, P., Pogosyan, A., and Brown, P. (2008). Patterns of bidirectional communication between cortex and basal ganglia during movement in patients with Parkinson disease. J. Neurosci. 28, 3008–3016.
Leblois, A., Meissner, W., Bioulac, B., Gross, C.E., Hansel, D., and Boraud, T. (2007). Late emergence of synchronized oscillatory activity in the pallidum during progressive parkinsonism. European Journal of Neuroscience 26, 1701–1713.
Levy, R., and Dubois, B. (2006). Apathy and the Functional Anatomy of the Prefrontal Cortex–Basal Ganglia Circuits. Cereb. Cortex 16, 916–928.
Levy, R., Hutchison, W.D., Lozano, A.M., and Dostrovsky, J.O. (2000). High-frequency Synchronization of Neuronal Activity in the Subthalamic Nucleus of Parkinsonian Patients with Limb Tremor. J. Neurosci. 20, 7766–7775.
Lewis, S.J.G., Foltynie, T., Blackwell, A.D., Robbins, T.W., Owen, A.M., and Barker, R.A. (2005). Heterogeneity of Parkinson’s disease in the early clinical stages using a data driven approach. J. Neurol. Neurosurg. Psychiatr. 76, 343–348.
Limousin, P., Krack, P., Pollak, P., Benazzouz, A., Ardouin, C., Hoffmann, D., and Benabid, A.-L. (1998). Electrical Stimulation of the Subthalamic Nucleus in Advanced Parkinson’s
Disease. New England Journal of Medicine 339, 1105–1111.
Litvak, V., Eusebio, A., Jha, A., Oostenveld, R., Barnes, G., Foltynie, T., Limousin, P., Zrinzo, L., Hariz, M., Friston, K., et al. (2012). Movement-Related Changes in Local and Long-Range Synchronization in Parkinson’s Disease Revealed by Simultaneous
Magnetoencephalography and Intracranial Recordings. J. Neurosci. 32, 10541–10553.
Litvak, V., Eusebio, A., Jha, A., Oostenveld, R., Barnes, G.R., Penny, W.D., Zrinzo, L., Hariz, M.I., Limousin, P., Friston, K.J., et al. (2010). Optimized beamforming for simultaneous MEG and intracranial local field potential recordings in deep brain stimulation patients. Neuroimage 50, 1578–1588.
Litvak, V., Jha, A., Eusebio, A., Oostenveld, R., Foltynie, T., Limousin, P., Zrinzo, L., Hariz, M.I., Friston, K., and Brown, P. (2011). Resting oscillatory cortico-subthalamic connectivity in patients with Parkinson’s disease. Brain 134, 359–374.
López-Azcárate, J., Tainta, M., Rodríguez-Oroz, M.C., Valencia, M., González, R., Guridi, J., Iriarte, J., Obeso, J.A., Artieda, J., and Alegre, M. (2010). Coupling between Beta and High-Frequency Activity in the Human Subthalamic Nucleus May Be a Pathophysiological Mechanism in Parkinson’s Disease. J. Neurosci. 30, 6667–6677.
45 45References
Lozano, A.., Lang, A.., Galvez-Jimenez, N., Miyasaki, J., Duff, J., Hutchison, W.., and Dostrovsky, J.. (1995). Effect of GPi pallidotomy on motor function in Parkinson’s
disease. The Lancet 346, 1383–1387.
Magariños-Ascone, C.M., Figueiras-Mendez, R., Riva-Meana, C., and Córdoba-Fernández, A. (2000). Subthalamic neuron activity related to tremor and movement in Parkinson’s
disease. European Journal of Neuroscience 12, 2597–2607.
Maris, E., Schoffelen, J.-M., and Fries, P. (2007). Nonparametric statistical testing of coherence differences. Journal of Neuroscience Methods 163, 161–175.
McAuley, J.H., and Marsden, C.D. (2000). Physiological and pathological tremors and rhythmic central motor control. Brain 123 ( Pt 8), 1545–1567.
Mink, J.W. (1996). The basal ganglia: focused selection and inhibition of competing motor programs. Prog. Neurobiol. 50, 381–425.
Mure, H., Hirano, S., Tang, C.C., Isaias, I.U., Antonini, A., Ma, Y., Dhawan, V., and Eidelberg, D. (2011). Parkinson’s disease tremor-related metabolic network: Characterization, progression, and treatment effects. NeuroImage 54, 1244–1253.
Muthuraman, M., Heute, U., Arning, K., Anwar, A.R., Elble, R.J., Deuschl, G., and Raethjen, J. (2012). Oscillating central motor networks in pathological tremors and voluntary movements. What makes the difference? NeuroImage 60, 1331–1339.
Nambu, A., Takada, M., Inase, M., and Tokuno, H. (1996). Dual somatotopical representations in the primate subthalamic nucleus: evidence for ordered but reversed body-map transformations from the primary motor cortex and the supplementary motor area. J. Neurosci. 16, 2671–2683.
Nolte, G., Bai, O., Wheaton, L., Mari, Z., Vorbach, S., and Hallett, M. (2004). Identifying true brain interaction from EEG data using the imaginary part of coherency. Clin Neurophysiol 115, 2292–2307.
Obeso, J.A., Rodriguez-Oroz, M.C., Goetz, C.G., Marin, C., Kordower, J.H., Rodriguez, M., Hirsch, E.C., Farrer, M., Schapira, A.H.V., and Halliday, G. (2010). Missing pieces in the Parkinson’s disease puzzle. Nature Medicine 16, 653–661.
Oswal, A., Brown, P., and Litvak, V. (2012). Movement related dynamics of subthalmo-cortical alpha connectivity in Parkinson’s disease. Neuroimage 70C, 132–142.
Özkurt, T.E., Butz, M., Homburger, M., Elben, S., Vesper, J., Wojtecki, L., and Schnitzler, A. (2011). High frequency oscillations in the subthalamic nucleus: a neurophysiological marker of the motor state in Parkinson’s disease. Exp. Neurol. 229, 324–331.
Pell, M.D., and Leonard, C.L. (2003). Processing emotional tone from speech in Parkinson’s disease: A role for the basal ganglia. Cognitive, Affective, & Behavioral
Neuroscience 3, 275–288.
Pogosyan, A., Gaynor, L.D., Eusebio, A., and Brown, P. (2009). Boosting Cortical Activity at Beta-Band Frequencies Slows Movement in Humans. Current Biology 19, 1637–1641.
46 46References
Pollok, B., Gross, J., Dirks, M., Timmermann, L., and Schnitzler, A. (2004). The cerebral oscillatory network of voluntary tremor. J. Physiol. (Lond.) 554, 871–878.
Rosin, B., Slovik, M., Mitelman, R., Rivlin-Etzion, M., Haber, S.N., Israel, Z., Vaadia, E., and Bergman, H. (2011). Closed-Loop Deep Brain Stimulation Is Superior in Ameliorating Parkinsonism. Neuron 72, 370–384.
Salenius, S., Avikainen, S., Kaakkola, S., Hari, R., and Brown, P. (2002). Defective cortical drive to muscle in Parkinson’s disease and its improvement with levodopa. Brain 125, 491–500.
Santa, W., Jensen, R., Miesel, K., Carlson, D., Avestruz, A., Molnar, G., and Denison, T. (2008). Radios for the brain? a practical micropower sensing and algorithm architecture for neurostimulators. In IEEE International Symposium on Circuits and Systems, 2008. ISCAS 2008, pp. 348 –351.
Schnitzler, A., and Gross, J. (2005). Normal and pathological oscillatory communication in the brain. Nat. Rev. Neurosci. 6, 285–296.
Schnitzler, A., and Hirschmann, J. (2012). Magnetoencephalography and neuromodulation. Int. Rev. Neurobiol. 107, 121–136.
Schnitzler, A., Timmermann, L., and Gross, J. (2006). Physiological and pathological oscillatory networks in the human motor system. J. Physiol. Paris 99, 3–7.
Selikhova, M., Williams, D.R., Kempster, P.A., Holton, J.L., Revesz, T., and Lees, A.J. (2009). A clinico-pathological study of subtypes in Parkinson’s disease. Brain 132, 2947–2957.
Sharott, A., Magill, P.J., Harnack, D., Kupsch, A., Meissner, W., and Brown, P. (2005). Dopamine depletion increases the power and coherence of beta-oscillations in the cerebral cortex and subthalamic nucleus of the awake rat. Eur. J. Neurosci. 21, 1413–
1422.
Syed, E.C.J., Benazzouz, A., Taillade, M., Baufreton, J., Champeaux, K., Falgairolle, M., Bioulac, B., Gross, C.E., and Boraud, T. (2012). Oscillatory entrainment of subthalamic nucleus neurons and behavioural consequences in rodents and primates. European Journal of Neuroscience 36, 3246–3257.
Timmermann, L., Gross, J., Dirks, M., Volkmann, J., Freund, H.-J., and Schnitzler, A. (2003). The cerebral oscillatory network of parkinsonian resting tremor. Brain 126, 199–212.
Turner, R.S., and Anderson, M.E. (1997). Pallidal discharge related to the kinematics of reaching movements in two dimensions. J. Neurophysiol. 77, 1051–1074.
Varela, F., Lachaux, J.-P., Rodriguez, E., and Martinerie, J. (2001). The brainweb: Phase synchronization and large-scale integration. Nat Rev Neurosci 2, 229–239.
Van Veen, B.D., and Buckley, K.M. (1988). Beamforming: a versatile approach to spatial filtering. IEEE ASSP Magazine 5, 4 –24.
47 47References
Wang, S., Aziz, T.Z., Stein, J.F., and Liu, X. (2005). Time–frequency analysis of transient neuromuscular events: dynamic changes in activity of the subthalamic nucleus and forearm muscles related to the intermittent resting tremor. Journal of Neuroscience Methods 145, 151–158.
Weinstein, S., and Ebert, P. (1971). Data Transmission by Frequency-Division Multiplexing Using the Discrete Fourier Transform. IEEE Transactions on Communication Technology 19, 628–634.
Whitmer, D., de Solages, C., Hill, B., Yu, H., Henderson, J.M., and Bronte-Stewart, H. (2012). High frequency deep brain stimulation attenuates subthalamic and cortical rhythms in Parkinson’s disease. Front Hum Neurosci 6, 155.
Wichmann, T., Bergman, H., Starr, P.A., Subramanian, T., Watts, R.L., and DeLong, M.R. (1999). Comparison of MPTP-induced changes in spontaneous neuronal discharge in the internal pallidal segment and in the substantia nigra pars reticulata in primates. Exp Brain Res 125, 397–409.
Wichmann, T., and Dostrovsky, J.O. (2011). Pathological basal ganglia activity in movement disorders. Neuroscience 198, 232–244.
Williams, D., Tijssen, M., Van Bruggen, G., Bosch, A., Insola, A., Di Lazzaro, V., Mazzone, P., Oliviero, A., Quartarone, A., Speelman, H., et al. (2002). Dopamine-dependent changes in the functional connectivity between basal ganglia and cerebral cortex in humans. Brain 125, 1558–1569.
48 48Erklärung
10 Erklärung
Hiermit erkläre ich, dass ich die vorgelegte Dissertation eigenständig und ohne unerlaubte
Hilfe angefertigt habe. Die Dissertation wurde in der vorliegenden oder in ähnlicher
Form noch bei keiner anderen Institution eingereicht. Ich habe bisher keine erfolglosen
Promotionsversuche unternommen.
Düsseldorf, den
Jan Hirschmann
49 49Danksagung
11 Danksagung
Mein besonderer Dank gilt meinem Doktorvater Prof. Dr. Alfons Schnitzler für seine
kontinuierliche Unterstützung, die ausgezeichnete wissenschaftliche Betreuung und
seine anhaltende und ansteckende Begeisterung für dieses Forschungsprojekt.
Prof. Dr. Tobias Kalenscher danke ich herzlich für die Zweitbetreuung dieser
Doktorarbeit.
Des Weiteren bin ich dem MEG-DBS Team ausgesprochen dankbar für die
hervorragende Zusammenarbeit und die große Hilfsbereitschaft, die diese Gruppe
auszeichnet. Dazu zählen Dr. Lars Wojtecki, Dr. Markus Butz, Dr. Tolga Özkurt, Dr.
Christian Hartmann, Dipl.-Psych. Saskia Elben, Dr. Nienke Hoogenboom und cand. med.
Melanie Homburger. Zudem danke ich Prof. Dr. Jan Vesper, der die Experimente von
neurochirurgischer Seite stets unterstützt hat.
Ohne die Hilfsbereitschaft der Patienten, die allesamt ohne eigenen Nutzen an den
teilweise langwierigen Messungen teilgenommen haben, wäre diese Arbeit nicht
möglich gewesen. Herzlichen Dank!
Für das Korrekturlesen dieser Arbeit gilt mein Dank Dr. Elisabeth May, Dr. Joachim
Lange und Dr. Markus Butz.
Zu guter Letzt möchte ich mich bei meinen Eltern und bei meiner Freundin Tina
bedanken, mit der ich alle Höhen und Tiefen der Promotion teilen konnte.
50 50Appendix
12 Appendix
This work is based on:
Appendix 1:
Hirschmann, J., Özkurt, T.E., Butz, M., Homburger, M., Elben, S., Hartmann, C.J., Vesper, J.,
Wojtecki, L., and Schnitzler, A. (2011). Distinct oscillatory STN-cortical loops revealed by
simultaneous MEG and local field potential recordings in patients with Parkinson’s
disease. Neuroimage 55, 1159–1168.
Impact factor (2011): 5.89
Personal contribution: 80%
Appendix 2:
Hirschmann, J., Özkurt, T.E., Butz, M., Homburger, M., Elben, S., Hartmann, C.J., Vesper, J.,
Wojtecki, L., and Schnitzler, A. (2013). Differential modulation of STN-cortical and
cortico-muscular coherence by movement and levodopa in Parkinson’s disease.
NeuroImage 68, 203–213.
Impact factor (2011): 5.89
Personal contribution: 80%
Appendix 3:
Hirschmann, J., Hartmann, C.J., Butz, M., Hoogenboom, N., Özkurt, T.E., Elben, S., Vesper,
J., Wojtecki, L., and Schnitzler, A. (2013). A direct relationship between oscillatory STN-
cortex coupling and rest tremor in Parkinson’s disease. Brain 136 (12), 3659-3670.
Impact factor (2012): 9.92
Personal contribution: 80%
51 51Appendix
Other aspects are taken from:
Appendix 4:
Schnitzler, A., and Hirschmann, J. (2012). Magnetoencephalography and
neuromodulation. Int. Rev. Neurobiol. 107, 121–136.
Impact factor (2011): 2.35
Personal contribution: 50%
Distinct oscillatory STN-cortical loops revealed by simultaneous MEG and local fieldpotential recordings in patients with Parkinson's disease
J. Hirschmann a,b, T.E. Özkurt a,b, M. Butz a,b, M. Homburger a,b, S. Elben a,b, C.J. Hartmann a,b, J. Vesper c,L. Wojtecki a,b, A. Schnitzler a,b,⁎a Institute of Clinical Neuroscience and Medical Psychology, Medical Faculty, Heinrich-Heine-University Düsseldorf, Düsseldorf, Germanyb Department of Neurology, University Hospital Düsseldorf, Düsseldorf, Germanyc Department of Functional Neurosurgery and Stereotaxy, University Hospital Düsseldorf, Düsseldorf, Germany
a b s t r a c ta r t i c l e i n f o
Article history:
Received 20 September 2010Revised 2 November 2010Accepted 19 November 2010Available online 29 November 2010
Keywords:
Parkinson's diseaseMagnetoencephalographySubthalamic nucleusFunctional connectivityOscillationsCoherence
Neuronal oscillations are assumed to play a pivotal role in the pathophysiology of Parkinson's disease (PD).Neurons in the subthalamic nucleus (STN) generate oscillations which are coupled to rhythmic populationactivity both in other basal ganglia nuclei and cortical areas.In order to localize these cortical areas, we recorded local field potentials (LFPs) and magnetoencephalography(MEG) simultaneously in PD patients undergoing surgery for deep brain stimulation (DBS). Patients werewithdrawn from antiparkinsonian medication and recorded at rest. We scanned the entire brain for oscillationscoherent with LFPs recorded from the STN with a frequency domain beamformer.Coherent activity in the low (12–20 Hz) and high (20–35 Hz) beta range was found in the ipsilateralsensorimotor and the premotor cortex. Coherence in the alpha range (7–12 Hz) was observed at variouslocations in the ipsilateral temporal lobe. In a subset of subjects, the superior temporal gyrus consistentlyshowed coherent alpha oscillations.Our findings provide new insights into patterns of frequency-specific functional connectivity between basalganglia and cortex and suggest that simultaneous inter-regional interactionsmay be segregated in the frequencydomain. Furthermore, they demonstrate that simultaneous MEG-LFP recordings are a powerful tool to studyinteractions between brain areas in PD patients undergoing surgery for DBS.
© 2010 Elsevier Inc. All rights reserved.
Introduction
Recordings from the basal ganglia of patients with Parkinson'sdisease (PD) undergoing surgery for deep brain stimulation (DBS)revealed strong oscillatory power in the alpha (7–12 Hz) and beta(12–35 Hz) band (Brown et al., 2001; Kühn et al., 2004; Levy et al.,2002; Priori et al., 2004). Furthermore, basal ganglia oscillations werefound to be coupled to oscillations in distant brain regions. Bysimultaneously recording electroencephalography (EEG) and localfield potentials (LFPs) it was shown that oscillations recorded fromthe STN are coherent with oscillations in cortical areas (Cassidy et al.,2002; Fogelson et al., 2006; Lalo et al., 2008; Marsden et al., 2001;Williams et al., 2002). Much like beta power in the STN, coherence inthe range from 10 to 30 Hz was found to be attenuated by movement(Cassidy et al., 2002; Lalo et al., 2008), the administration of levodopa(Lalo et al., 2008; Williams et al., 2002) and DBS (Kühn et al., 2008).
Although these findings suggest that abnormal coupling betweenSTN and cortical oscillations may be pathophysiologically relevant,
the cortical areas engaged in this coupling have not been identified sofar. Simultaneous EEG-LFP recordings provided first evidence that thedistribution of coherence across cortical areas is heterogeneous andfrequency-dependent (Fogelson et al., 2006; Williams et al., 2002).However, the exact topography of STN-cortical coherence remains tobe determined.
In this study we utilized simultaneous magnetoencephalography(MEG)-LFP recordings to map STN-cortical coherence. In contrast toEEG, MEG allows for whole-head, post-surgical measurements andthus for source localization with high spatial resolution. Using afrequency domain beamformer (Gross et al., 2001), we localized STN-cortical coherence in eight PD patients. While the feasibility of thisapproach has recently been demonstrated with data from a singlesubject (Litvak et al., 2010), it has not been realized in a group ofpatients so far.
Materials and methods
Patients
Nine patients (three females) with idiopathic, akinetic-rigid PD(mean age: 64±7.6 years, range: 47–75), who were clinically selected
NeuroImage 55 (2011) 1159–1168
⁎ Corresponding author. Institute of Clinical Neuroscience and Medical Psychology,Heinrich-Heine-University Düsseldorf, Universitätsstr, 1, 40225 Düsseldorf, Germany.
E-mail address: [email protected] (A. Schnitzler).
1053-8119/$ – see front matter © 2010 Elsevier Inc. All rights reserved.doi:10.1016/j.neuroimage.2010.11.063
Contents lists available at ScienceDirect
NeuroImage
j ourna l homepage: www.e lsev ie r.com/ locate /yn img
for DBS of the STN, participated in the study. One patient was excludeddue to severe head movement artifacts. All patients gave writteninformed consent. Table 1 summarizes the clinical details.
The study was approved by the local ethics committee (study no.3209) and is in accordance with the Declaration of Helsinki. Highresolution T1-weighted magnetic resonance (MR) images wereobtained from each patient prior to surgery. Antiparkinsonianmedication could be substantially reduced in all but two subjects,for whom the levodopa equivalent dose (LED) stayed approximatelystable. The average reduction in LED was 40%±26%.
Planning and implantation
All oral antiparkinsonian medication was withdrawn the evening
before surgery and substituted by subcutaneous apomorphine
medication using a medication pump.
Electrode implantation was performed at the Department of
Stereotaxy and Functional Neurosurgery in Düsseldorf. Patients
were bilaterally implanted during medication OFF.
The STN was targeted on the basis of Schaltenbrand–Wahren atlas
coordinates (Schaltenbrand and Wahren, 1977), using stereotactic
cranial computer tomography (CT) and high resolution MRI. We
performed intra-operative microelectrode recordings using the
INOMED MER system (INOMED Inc., Tenningen, Germany) to
determine the STN borders and the optimal implantation area.
Intraoperative recordings were performed with up to five microelec-
trodes. The anterior, posterior, lateral and medial microelectrodes
were concentrically configured around the central electrode, each
with a distance of 2 mm from the central electrode (Ben's gun
system). The final placement of the DBS electrode (model 3389,
Medtronic Corporation, Minneapolis, MN, USA) was based on multi-
unit activity and a clinical profile of stimulation effects and side
effects.
Following surgery, the locations of the four DBS electrode contacts
were derived from postoperative stereotactic CT images. Fig. 1
illustrates the average position of the contacts yielding the best
clinical effect when used for chronic DBS. Table S1 lists the stereotactic
coordinates of all contacts for all patients.
Recordings
All recordings were performed with a 306-channel, whole-head
MEG system (Elekta Oy, Helsinki, Finland). Recording sessions took
place the day after electrode implantation. At this stage, the
stimulator was not yet implanted so that the externalized leads of
the intracranial electrodes could be connected to the EEG acquisition
device of our system. We used non-magnetic extension leads
(Medtronic Bakken Research Center, Maastricht, the Netherlands)
for electrode connection, thereby avoiding the strong artifact induced
by ferromagnetic connectors (Litvak et al., 2010).
Administration of apomorphine was discontinued at least 2 h prior
to recordings so that all patients were in a defined medication OFF
during recordings. We verified the OFF state by performing ratings
according to the motor section of the Unified Parkinson's Disease
Rating Scale (UPDRS III; Goetz et al., 2008).
We recorded MEG, LFPs, vertical and horizontal electro-oculo-
grams (EOGs) and the electromyogram (EMG) of the extensor and
flexor muscles of both forearms simultaneously. DBS Electrode
contacts were referenced against a surface electrode at the left
mastoid and rearranged into a bipolar montage offline. EMG was
measured with reference to surface electrodes at the forearm tendons.
Sampling rate was 2 kHz. MEG signals were band-pass filtered online
between 0.03 and 660 Hz. LFP and EMG signals were filtered between
0.1 and 660 Hz. Patients were recorded at rest for 5 min with no
specific task, but instructed to relax and move as little as possible.
Preprocessing and artifact removal
EMG data were visually inspected and periods containing move-
ment were discarded. A notch filter at 50 Hz was applied to MEG and
LFP data to remove power line noise. TheMaxFilter software (Elekta Oy,
Helsinki, Finland) was used to apply signal space separation (SSS) to
MEG data (Taulu and Kajola, 2005). The method served to remove
interferences arising from far outside the MEG shielded room and to
reconstruct data from noisy or dysfunctional channels (four channels
for each patient). The algorithm performs reconstruction by interpo-
lation, using spherical harmonic basis functions. Based on previous
work on the application of SSS (Ahonen et al., 1993; Nenonen et al.,
2007; Song et al., 2008; Taulu et al., 2005), we chose the expansion
limits of the spherical harmonic functions as Lin=8 for the inner
sources and Lout=3 for the outer sources.
Data analysis
Data were analyzed using custom-made Matlab (The Mathworks,
Natick, Massachusetts, USA) scripts, most of which were based on
Fieldtrip, a Matlab-based, open source analysis toolbox (http://
fieldtrip.fcdonders.nl/). We only analyzed data recorded by the 204
planar gradiometers of the MEG system.
Coherence has been widely used to quantify similarity between
neuronal oscillations and is commonly interpreted as interaction or
communication between brain areas (Fries, 2005; Schnitzler and
Gross, 2005). It may be understood as a measure of amplitude and
phase consistency across epochs of recorded data (Maris et al., 2007).
Localization of coherent sources was realized by Dynamic Imaging of
Coherent Sources (DICS), a frequency domain beamforming approach
Table 1
Clinical information on subjects. UPDRS motor scores (Goetz et al., 2008; sum score: 132) were obtained the day before (column 6) and the day after surgery (column 7) to quantify
the clinical effects of medication. Further UPDRS scores were obtained 3 months after surgery to quantify the clinical effect of DBS in the medication OFF (column 8). The contacts
chosen for DBS in clinical tests 3 months after surgery and their stereotactic coordinates relative to the mid-commissural point are given in columns 9–12. A single number in
columns 10 and 12 indicates that stimulation was monopolar, otherwise it was bipolar (+ = cathode, −=anode).
Patient Age
(years)
Sex Disease
duration
(years)
Predominant
side
Motor UPDRS
Med-OFF
preOP
Motor UPDRS
Med-OFF/Med-ON
postOP
Motor UPDRS
DBS-OFF/DBS-
ON
Stimulated
contacts
left
Coordinates
stimulated
contacts left
Stimulated
contacts right
Coordinates
stimulated
contacts right
BH 75 f 26 Left 35 50/32 42/29 2− [−12.5, 0.0, 1.4] 2− [10.8, −1.7, −0.1]
HH 69 m 11 Right 33 41/33 unavailable 1− [−14.2, −1.7, 0.01 1− [14.5, −0.1, 0.0]
KH 62 m 16 Right 36 44/30 52/28 1− [−13.1, −1.1, −1.4] 1− [13.8, −2.8, −0.9]
LW 70 m 11 Right 28 39/31 23/15 1− [−13.1, −2.8, −2.6] 2− [12.1, −2.0, −1.7]
PH 62 f 15 Left 43 27/21 46/37 3+ [−12.4, 0.7, 2.4] 3+ [14.1, 0.7, 2.5]
2− [−11.7, −0.9, 0.7] 2− [13.4, −0.9, 0.8]
SA 72 m 16 Left 44 43/38 29/13 1− [−12.9, 2.0, −2.3] 2− [12.7, 4.5, −2.4]
WA 64 f 18 Left 35 24/9 34/21 2+
1−
[−11.9, 2.1, −0.2] 2+ [14.6, 0.0, −1.9]
[−11.2, 0.4, −2.0] 1− [13.9, −1.7, −3.7]
1160 J. Hirschmann et al. / NeuroImage 55 (2011) 1159–1168
(Gross et al., 2001). Activity recorded in the four LFP channels(Fig. 2A) served as the reference signals for coherence computation ina dense grid of spatial locations spanning the entire brain. Coherencewas analyzed separately in four frequency bands: alpha (7–12 Hz),low beta (12–20 Hz), high beta (20–35 Hz) and gamma (70–90 Hz).Because DICS works frequency-wise, analyzing a range of frequenciesfor each patient and LFP channel would be extremely computationallydemanding. Therefore, we estimated the frequency of strongestcoherence prior to source localization based on MEG sensor data.For each patient, LFP channel and frequency band, we identified thefiveMEG sensors with the highest mean coherence and averaged theircoherence spectra. We automatically searched the averaged spectrafor the highest local maximum and chose its frequency for sourcelocalization. Fig. 2B and C illustrates the procedure.
DICS projects sensor data through a spatial filter derived from theleadfield and the cross-spectral densities (CSDs). To obtain a goodestimate of the latter, they were calculated for overlapping, Hanning-tapered segments of the continuous rest data (varying between 150and 300 s) and averaged over segments. The regularization parameterfor DICS was set to 0.5% of the mean power. Frequency resolution forDICS was 3.9 Hz. Low frequency resolution was necessary to provide
good CSD estimates and therewith reliable source localization.
The leadfield was calculated using a single-shell head model based
on the patients’ individual MRIs (Nolte, 2003). MRIs were first
normalized to the T1 template brain included in SPM2 (http://www.
fil.ion.ucl.ac.uk/spm/). The inverse of that normalization was then
applied to a beamformer grid based on the template brain (Mattout et
al., 2007). As a consequence, all grid points in the warped grid
corresponded to a grid point in the template grid and thus had a
defined position in Montreal Neurological Institute (MNI) space. Grid
spacing was 5 mm.
Since only a single spatial maximumwas usually discernible in the
functional images, we defined a source as the cluster of supra-
threshold voxels showing maximal coherence. The threshold was
defined according to Halliday et al. (1995). Sources were required to
consist of eight neighboring supra-threshold voxels at least. The
number 8 was chosen because simulations had shown that eight
neighbors occurred rarely (pb0.03) when randomly setting 5% of
voxels above threshold. We note that this procedure does not provide
rigorous control over the number of false positive detections but is
preferable to simple thresholding. A nonparametric significance test,
which effectively controls for multiple comparisons, was applied to
test for the significance of coherence on the group level (see Statistics
section).
For visualization, functional images were coregistered to a
canonical MR image, using nasion, left and right pre-auricular point,
and interpolated on the MR image. MNI coordinates of the maxima of
interpolated images were identified with the help of the AFNI atlas
(http://afni.nimh.nih.gov/afni), integrated into Fieldtrip.
Statistics
Statistical analyses were performed using Matlab. When compar-
ing LFP-MEG coupling across subjects or frequency bands we used the
Fig. 1. Average position of the contacts showing the best clinical effect when used for chronic DBS. Stereotactic contact coordinates were normalized with respect to the
distance between the anterior commissure (AC) and the posterior commissure (PC) and projected onto the Schaltenbrand–Wahren atlas (Schaltenbrand andWahren, 1977).
When stimulation was bipolar, the mean coordinates of the contact pair were used for averaging across subjects. Left: Average position of stimulated contacts (left
hemisphere: x=−12.4 mm±1.44 mm, z=−0.6 mm±1.55 mm; right hemisphere: x=13 mm±1.32 mm, z=−0.9 mm±1.44 mm) projected onto coronal slice 3 mm
behind mid-commissural point (MCP). The coordinates indicate that the points are situated behind MCP (negative y, given by slice) and below MCP (negative z). Right:
Average position of stimulated contacts (left hemisphere: x=−12.4 mm±1.44 mm, y=−0.7 mm±1.79 mm; right hemisphere: x=13 mm±1.32 mm, y=
−0.7 mm ±2.35 mm) projected onto axial slice 1.5 below AC-PC line. Grid spacing is 9 mm; yellow crosses depict standard deviations.
Fig. 2. Reference scheme for DBS electrodes and example for the choice of frequency. A) Schematic drawing of the DBS electrode. Electrode contacts were re-referenced offline,
resulting in four bipolar LFP channels. B) The five MEG sensors with the highest average coherence in the high beta band in one exemplary subject (Patient HH, LFP channel R03).
Note that the spatial neighborhood of these sensors is of physiological origin and not resulting from the application of any spatial criteria. C) Coherence averaged across the five
channels depicted in B). Frequency resolution is 1 Hz. The red line depicts the significance threshold. The green dot marks the highest peak. For this particular patient (HH), LFP
channel (R03) and frequency band (high beta) the frequency chosen for source localization was 28 Hz.
1161J. Hirschmann et al. / NeuroImage 55 (2011) 1159–1168
Fisher-transformed modulus of coherency instead of coherence(Brillinger, 1981). We will refer to this measure as z-transformedcoherence.
Significance of coherence on the group level was tested by anonparametric test (Nichols and Holmes, 2002; Schoffelen et al.,2008). The test corresponded to a one-sided, single sample t-test. Weconsidered only functional images showing maximal coherence for agiven subject and frequency band (grid spacing: 1 cm, frequencyresolution: 1.9 Hz). First, we computed an image of z-transformeddifference values for each subject (z=arctanh(sqrt(coh)) - arctanh(sqrt(threshold)), threshold according to Halliday et al. (1995)) andaveraged the difference images.We then determined themaximumofthe average image, which served as the test statistic. The nullhypothesis stated that the individual difference values are symmet-rically distributed around zero. Under this assumption, alternativesamples may be generated by flipping the sign of any subset ofdifference images prior to calculating the test statistic. The fullpermutation distribution of the test statistic was computed by flippingthe sign of all possible subsets and recalculating the test statistic. Thecritical value was defined as the 0.05 × N highest statistic of the
permutation distribution,N being the number of permutations. Voxels
showing z-values higher than the critical value were considered
significant.
For the assessment of the distribution of peak coherence across
electrode channels, we used a parametric test developed by Amjad et al.
(1997). The significance level was Dunn–Šidák corrected for multiple
comparisons.
Results
Sensor level analysis
We observed significant coherence between LFPs andMEG sensors
in the alpha, low beta and high beta band in all subjects (see Fig. S2 in
the supplementary material). Coherence lateralized to the ipsilateral
side with respect to the STN. On average, sensors ipsilateral to the STN
showed higher coherence than contralateral sensors (Fig. 3). As in the
example shown in Fig. 4, coherence was usually strongest in sensors
located above the paramedian central sulcus region.
Except for a single patient (patient ZE), significant coherence in
the gamma band was not observed. This patient was exceptional as he
had received DBS of the globus pallidus internus for several years
before another pair of electrodes was implanted for STN stimulation.
As coherence in the gamma band was seen only in this one instance,
we did not take the gamma band into account in subsequent analysis.
Source level analysis
Characterization of coherent sources
Having identified the frequencies at which coherence was highest
on the sensor level, we used DICS to localize coherent activity at these
frequencies. In the following, we will use the term “source” when
referring to areas of elevated STN-cortical coherence detected by the
procedure outlined in Data analysis section. We identified 161
sources, thereof 53 coherent in the alpha band, 48 in the low beta
band and 60 in the high beta band. Please note that with our definition
a source signifies localized coherent activity with respect to a given
LFP channel. In case signals of two LFP channels are similar, as it may
be the case for neighboring channels, two sources of similar frequency
and location may well reflect the same cerebral activity. As we
accepted only one source per frequency band and LFP channel, the
highest possible number of sources for a given frequency band was 64
(eight patients, eight channels).
Fig. 5 shows an alpha and a high beta source found for the same LFP
channel. The example demonstrates that a single recording site in the
basal ganglia may show coherence with different cortical areas at
different frequencies.
We investigated whether the strength of oscillatory coupling
differed across frequency bands. An analysis of variance (ANOVA)
revealed that z-transformed coherence at the source maximum and
mean source coherence (averaged across voxels) was similar for all
frequency bands (Fmax(2)=1.2, pmean=0.3; Fmean(2)=2.01,
pmean=0.14).
Topography of coherent sources
In order to investigate frequency-dependent differences in the
spatial topography of sources, we pooled all coherent sources
identified for a given frequency band and determined the MNI
coordinates of their maxima. The coordinates were then assigned to
brain regions with the help of the AFNI atlas.
The vast majority of source maxima was located ipsilateral to the
STN (alpha: 83%, low beta: 88%, high beta: 95%). As can be seen from
Fig. 6, there was a clear difference between alpha and beta sources
with respect to the locations of their maxima.
Alpha source maxima weremostly situated in the temporal cortex.
They did not show a consistent topographywhen pooled over subjects
and LFP channels but scattered across a large area. However, one
larger cluster was found in the posterior part of the superior temporal
gyrus (STG; 11 sources, 5 hemispheres, 4 subjects). For the three
patients showing unilateral alpha coherence in STG, coherence was
located on the left side, which was ipsilateral to the body side most
affected by PD symptoms in all three cases. Alpha coherence was also
found in the postcentral gyrus (PostCG; 6 sources, 3 hemispheres, 3
subjects) and precentral gyrus (PreCG; 4 sources, 4 hemispheres, 4
subjects).
Contrary to alpha sources, high beta sources clustered in one area.
In seven out of eight subjects we found high beta source maxima to
cluster in the ipsilateral sensorimotor cortex. The cluster encom-
passed PreCG (26 sources, 11 hemispheres, 6 subjects), PostCG (7
sources, 4 hemispheres, 4 subjects) and parts of Brodmann area 6
rostral to PreCG (5 sources, 3 hemispheres, 3 subjects). All source
maxima in PostCG were located contralateral to the body side most
affected by PD. In contrast, all source maxima in premotor areas were
situated ipsilateral to the latter.
The topography of low beta sources resembled that of high beta
sources, with source maxima clustering around PreCG (10 sources, 6
hemispheres, 5 subjects). We did not observe clear differences in the
topographies of low and high beta source maxima within cortical
motor areas. Low beta source maxima were also found in the fusiform
gyrus, PostCG, the insula, the temporal lobe, the cerebellum, and the
frontal lobe, but did not cluster in a specific region other than the
sensorimotor cortex.
Fig. 3. Lateralization of coherence on the sensor level. z-transformed MEG-LFP
coherence averaged across ipsilateral (i) and contralateral (c) MEG sensors. Values
for all subjects and LFP channels were pooled. Asterisks indicate p-values (**pb0.01,
***pb0.001, Wilcoxon rank sum test).
1162 J. Hirschmann et al. / NeuroImage 55 (2011) 1159–1168
A nonparametric statistical test revealed that subjects consistently
showed beta coherence in the ipsilateral sensorimotor cortex. High
and low beta coherence around the central sulcus was significantly
higher than threshold when averaged across subjects (Fig. 6D).
Coherence in the alpha band was not significant on the group level in
any brain area.
Finally, we tested whether sources coherent with different LFP
channels in the same frequency band differed in their x, y or z
coordinates in MNI space. One-way ANOVAs did not yield any
evidence for a systematic relationship between the sources’ coordi-
nates and the LFP channels sources were coherent with.
Distribution of coherence across LFP channels
The first part of the analysis revealed that beta oscillations in
PreCG and alpha oscillations in STG were coherent with LFPs recorded
from the STN. In the second part, we defined the spatial maxima of
sources found in these areas as regions of interest (mean MNI
coordinates STG:±51,−20, 5; PreCG: ±33,−22, 57) and computed
the full coherence spectrum with all LFP channels. If several sources
had been identified in the same PreCG or STG (relating to different LFP
channels of the same subject), we chose the source of strongest
coherence. The aim of this analysis was to investigate the spatial
distribution of coherence within the area recorded by the DBS
electrode.
As expected, the spectra of STG and PreCG sources showed clear
differences. Fig. 7 shows an example of the distribution of alpha
(Fig. 7A) and beta coherence (Fig. 7B) across LFP channels.
STG sources showed strongest coherence in the alpha band. The
frequency of peak coherence was 11.3 Hz on average. In all cases,
coherence with ipsilateral channels was markedly higher than with
contralateral channels. When considering only ipsilateral channels,
we found that the distribution of peak alpha coherence across LFP
channels was not significantly different from a homogeneous
distribution in any case (Table 2), suggesting that alpha oscillations
Fig. 4. Example of MEG-LFP coherence in the high beta band (Patient HH, LFP channel R03). The figure shows the coherence spectra of all planar gradiometers (array is seen from
above, the subject's nose points to the top of the page). Note that coherence peaks in medial sensors, ipsilateral to the STN. Frequency resolution is 2 Hz.
Fig. 5. An exemplary alpha (upper row) and high beta (lower row) source localized by DICS. Coherence is color-coded. Both sources are coherent with activity recorded in the same
LFP channel (Patient LW, LFP channel L12), but at different frequencies (6.8 and 23.4 Hz).
1163J. Hirschmann et al. / NeuroImage 55 (2011) 1159–1168
in STG are coupled to more widely distributed STN alpha rhythm.
Consistent with this idea, the LFP channel with the largest contact
spacing (channel 03) showed the strongest coherence on average.
PreCG sources showed the highest coherence with LFP channels in
the beta band. The average frequency of peak coherence was 22.6 Hz.
As in the example shown in Fig. 7B, PreCG-LFP coherence was focal in
most cases. The null hypothesis of equal peak beta coherence with all
ipsilateral LFP channels was rejected in 7 of 12 cases. Fig. 8 shows the
average distribution of peak beta coherence across LFP channels for
these seven cases. On average, coherence was strongest with the
ipsilateral channel 12 and weakest with channel 01. However, the
difference between channels was not significant (F(3)=1.27, p=0.3).
Discussion
This study shows that the ipsilateral sensorimotor and adjacent
premotor cortex is the main source of cortical activity coherent with
beta oscillations in the STN of PD patients. Moreover, it identified
ipsilateral temporal areas as a source of coherent alpha activity.
Methodological considerations
Before discussing the physiological aspects of the results in detail,
we will make some methodological considerations. Coherence
analysis of electrophysiological recordings bears the risk to overes-
timate or falsely detect oscillatory coupling due to field spread
(Schoffelen and Gross, 2009; Srinivasan et al., 2007; Winter et al.,
2007).When the same signal is picked up at separate sites of
measurement, this will lead to the detection of coherence even if
the local activities at these sites are uncorrelated. There are several
reasons to believe that the results reported here are not due to field
spread.
First, we used beamforming to compute coherence, a technique
designed to suppress signals which do not originate from the location
that is being scanned (Van Veen et al., 1997). Beamforming is known
to alleviate the effects of field spread (Schoffelen and Gross, 2009).
Second, we used bipolar referencing of electrode contacts, i.e. the
LFP channels recorded local activity. In principle, it is conceivable that
such local activity was recorded byMEG sensors as well, despite its low
Fig. 6. Frequency-dependent differences in the topography of coherence. Colored squares represent the maxima of all alpha (A), low beta (B) and high beta (C) sources detected in
this study. Maximawere enlarged for visualization and interpolated on a canonical MR image. Sources coherent to the left STN have beenmirrored into the contralateral hemisphere.
LFP channels with which sources were coherent are color-coded. D) z-transformed coherence averaged across subjects. Only significant values (α=0.05, corrected) are displayed.
1164 J. Hirschmann et al. / NeuroImage 55 (2011) 1159–1168
amplitude. However, signals from sources as deep as the basal ganglia
would be detected by a large number of sensors and the amplitude
distribution across sensors would be roughly uniform. In this case,
coherent sources would be mapped to deep, central areas by
beamforming. Signals reaching MEG sensors from deep areas by field
spread are not expected to localize to cortical areas. Furthermore,
coherence induced by field spread is not expected to be physiologically
plausible, i.e. field spread cannot explain why beta coherence is
especially high in motor areas.
Apart from conduction of neuronal signals, artifacts may lead to
erroneous results in connectivity analysis (Schoffelen and Gross,
2009). Artifact handling is especially challenging in simultaneous
MEG-LFP recordings, as ferromagnetic components of the recording
setup may severely impair data quality (Litvak et al., 2010). In this
study we used exclusively non-magnetic hardware. As a result, the
data quality obtained in this study was comparable to the quality
obtained when the MEG signal is recorded from subjects without any
implants. To improve data quality further, we removed signal
Fig. 7. Examples of the distribution of coherence across electrode contacts. The blue lines depict coherence and the dotted, black lines the significance threshold. Labels of LFP
channels are shown to the right of the coherence spectra. Frequency resolution is 2 Hz, and frequency smoothing due to multi-tapering is 3.9 Hz. A) STG-LFP coherence (patient PH).
Scaling is the same in all subplots. B) PreCG-LFP coherence (patient LW). Scaling is the same in all subplots. Note that the distribution of coherence across LFP channels is roughly
homogeneous for STG signals, whereas it is focal for PreCG signals.
Table 2
Coherence between cortical signals and LFP channels. The fourth column contains the
probabilities that peak band coherence is the same with all LFP channels. The fifth
column contains the probabilities that peak band coherence is the same with all
ipsilateral LFP channels. Probabilities below significance level (α=0.05, corrected) are
marked by bold print. All channels listed in the second column are ipsilateral channels.
Brain area Channel with
highest peak
Frequency of
peak (Hz)
p value
(all channels)
p value
(ipsilateral channels)
STG 03 9.8 b0.001 0.91
STG 03 11.7 0.14 0.47
STG 03 11.7 0.21 0.25
STG 03 11.7 0.007 0.18
STG 01 11.7 0.004 0.1
PreCG 03 13.6 b0.001 b0.001
PreCG 12 21.5 b0.001 b0.001
PreCG 23 15.6 0.03 0.016
PreCG 12 29.3 b0.001 b0.001
PreCG 03 23.4 b0.001 b0.001
PreCG 01 29.3 0.062 0.2
PreCG 03 25.3 b0.001 0.11
PreCG 23 27.3 0.0015 0.027
PreCG 23 21.5 b0.001 0.013
PreCG 23 19.5 b0.001 b0.001
PreCG 03 25.4 0.058 0.83
PreCG 03 19.5 b0.001 b0.001
Fig. 8. Average distribution of peak beta coherence across electrode contacts
(i=ipsilateral, c=contralateral). For each hemisphere, we searched the eight
coherence spectra for the highest peak and determined the frequency at which it
occurred. Subsequently, we normalized coherence in each channel at that frequency by
dividing by the peak value. Normalized coherences were then averaged across
hemispheres. Bars depict standard deviations.
1165J. Hirschmann et al. / NeuroImage 55 (2011) 1159–1168
components attributable to sources outside the head by the
application of SSS (Taulu and Kajola, 2005).
Finally, transient tissue responses to electrode insertion, such as
edema, may be a methodological concern. LFP power has been shown
to be variable in the first days after surgery, and different frequency
bands may be differentially affected by transient changes in the tissue
(Rosa et al., 2010). Accordingly, coherence measured at later stages
may in principle differ from the data presented in this study. However,
we do not expect our main results to be qualitatively affected by
edema, since coherent sources may remain undetected due to edema,
but edema are unlikely to introduce significant coherences or a
consistent remapping of coherent sources
STN-cortical coupling
Using simultaneous MEG-LFP recordings in PD patients, we found
cortical activity to be coherent with STN LFPs in the alpha and beta, but
not in the gamma band. This finding is in agreementwith simultaneous
EEG-LFP recordings of PD patients in the medication OFF (Williams et
al., 2002). While significant coherence in the alpha and beta band was
readily observed after withdrawal of antiparkinsonian medication,
coherence in the gamma band emerged only after the administration of
levodopa (Lalo et al., 2008; Williams et al., 2002).
Coherence in the high beta band
Owing to the large number of sensors available in MEG recordings,
we were able to localize coherent activity on the source level. We
found coherent high beta activity to be generated ipsilaterally in
PreCG, more rostral sites in Brodmann area 6 and in PostCG.
Functionally, these areas represent the primary motor cortex (M1),
the premotor cortex and the primary somatosensory cortex, respec-
tively.We thus conclude that the area of strongest oscillatory coupling
in the high beta band is the ipsilateral sensorimotor and adjacent
premotor cortex.
Anatomically, a functional interaction between the motor cortex
and the STN is plausible. Both primary and secondary motor cortex
have abundant connections to the STN (Alexander et al., 1986;
Alexander and Crutcher, 1990). Cortical inputs may reach the STN via
the striatum and the external segment of the globus pallidus.
Alternatively, they are transmitted without relay via the so-called
hyperdirect pathway (Nambu et al., 1996). In turn, subthalamic
output reaches the motor cortex via the ventrolateral thalamus
(Hoover and Strick, 1999).
As both the motor cortex and the STN are somatotopically
organized (Rodriguez-Oroz et al., 2001) and regions representing
the same body part are anatomically connected (Miyachi et al., 2006;
Nambu et al., 1996), it is conceivable that corresponding regions in the
STN and the motor cortex show selective functional coupling. We
assessed this question by investigating whether the position of a
coherent cortical source varies systematically with the LFP channel
used as reference for coherence calculation. The absence of such a
relationship in our data suggests that, if selective functional coupling
occurs, it may not be readily detectable in post-surgical recordings
with DBS electrodes. The contacts of DBS electrodes are large
compared to the scale of somatotopic representation in the STN and
the direction of electrode insertion (mainly dorsoventral) does not
match the direction of somatotopic representation (mainly medio-
lateral). Moreover, inter-individual differences in the exact position of
DBS electrodes and the presence of edema might have hindered the
detection of a significant relationship.
Interhemispheric differences in STN-cortical beta coherence
Interestingly, this study reveals first indications for a hemispheric
asymmetry of STN-cortical beta coherence, reminiscent of the motor
symptom asymmetry typical for PD (Djaldetti et al., 2006). High beta
source maxima were found in the PostCG contralateral to the body
side most affected by PD, but not in the ipsilateral PostCG. In contrast,
high beta coherence in premotor areas occurred only ipsilaterally.
Thus, the patterns of STN-cortical coherence may not be identical for
both hemispheres. The hemisphere more affected by dopamine
depletion, i.e. contralateral to the more affected body side, may
show coherence with a more posterior sensorimotor area than the
less affected hemisphere. Note, however, that the data presented here
provide only indirect evidence, as the observed difference between
hemispheres is based on an across-subject comparison.
Coherence in the low beta band
We subdivided the beta band into low and high beta since there is
evidence for the existence of two functionally distinct beta rhythms in
the STN (Marceglia et al., 2006; Priori et al., 2004). One may speculate
that the underlying thalamo-cortical loops are spatially segregated. In
this study, neither the automatic detection of maximally coherent
sources nor the inspection of unprocessed functional images yielded
any evidence for a difference in topography. Our results therefore
suggest that the proposed difference in function relies on distinct
coupling frequencies rather than on interactions with spatially
distinct cortical areas. We note, however, that frequency resolution
was kept rather low in this study for the sake of precise source
localization. Low frequency resolution might have precluded the
detection of subtle differences between the topographies of low and
high beta coherence.
The occurrence of low beta sources in both motor and temporal
areas suggests that defining frequency bands according to the
traditional EEG literature may not lead to an optimal spatial
separation of cortical coherent sources. Choosing a higher frequency
as the border between the alpha and the low beta band would most
probably have resulted in clearer spatial separation of sources. With a
different definition of frequency bands, it is conceivable that sources
coherent in the extended alpha band would localize to the temporal
lobe, while sources coherent in the truncated beta band cluster in
cortical motor areas.
Apart from the choice of frequency band width, non-linear
correlations between alpha, low and high beta oscillations in the
STN may have contributed to the co-localization of cortical coherent
sources (Marceglia et al., 2006).
Distribution of beta coherence across LFP channels
Analysis of the distribution of spectral peaks across LFP channels
revealed that beta coherence with PreCG was not spatially uniform
but showed relative focality within the recording range of the DBS
electrode. This finding is in line with reports on intra-operative LFP
recordings showing that beta power depends on LFP recording site
within the STN area (Kühn et al., 2005; Trottenberg et al., 2007;
Weinberger et al., 2006).
Since post-surgical imaging does not allow for contact localization
with mm precision, the exact site of elevated beta coherence within
the STN area could not be determined. We found the LFP channel
showing maximal PreCG-LFP beta coherence to vary substantially
across hemispheres, most likely due to differences in electrode
placement. Accordingly, there was no significant difference across
LFP channels on average. However, the average distribution of beta
coherence showed an increase from the ventralmost (channel 01) to
more dorsal LFP channels (Fig. 8). It is therefore reasonable to assume
that PreCG-LFP beta coherence is increased in the dorsal part of the
STN, which constitutes the motor part of the nucleus (Rodriguez-Oroz
et al., 2001). Beta power has been shown to be locally elevated in this
area (Trottenberg et al., 2007).
Coherence in the alpha band
We localized coherent alpha activity in the ipsilateral temporal
cortex. This new finding contradicts previous studies investigating
EEG-LFP coherence, which localized alpha coherence to mesial rather
1166 J. Hirschmann et al. / NeuroImage 55 (2011) 1159–1168
than lateral sensors (Fogelson et al., 2006; Williams et al., 2002). A
reason for the divergent localizations could be that those previous
studies included tremor-dominant patients while we included
exclusively PD patients of the akinetic-rigid subtype. Parkinsonian
rest tremor is characterized by strong coherence in the alpha range
between activity in cortical motor areas, muscle activity and activity
in several other brain regions, including deep, diencephalic structures
(Pollok et al., 2009; Timmermann et al., 2003). Thus, alpha coherence
due to tremor may have dominated the spatial patterns of STN-
cortical alpha coherence described earlier. Previous studies may also
have missed STN-temporal lobe coupling due the limited spatial
sampling of cortical signals in post-surgical EEG recordings.
While alpha sources showed substantial spatial variability across
subjects and electrode contacts, a subset of subjects consistently
showed alpha coherence in the ipsilateral STG. In line with the diffuse
distribution of alpha power in the subthalamic region (Kühn et al.,
2005), the pattern of alpha coherence suggests that activity in STG is
coherent to a rather global alpha rhythm. Thus, STG-LFP coherence
does not necessarily reflect a specific interaction between STG and
STN but may well relate to other basal ganglia structures.
Anatomically, an interaction between the temporal cortex and the
basal ganglia could be explained by temporal projections to the
ventromedial caudate nucleus (Selemon and Goldman-Rakic, 1985;
Van Hoesen et al., 1981; Yeterian and Van Hoesen, 1978) or by
projections from the substantia nigra pars reticulata to the inferior
temporal cortex (Middleton and Strick, 1996). Functionally, STG has
been ascribed a role in perception rather than action. It is involved in
auditory (Buchsbaum et al., 2001; Howard et al., 2000) movement
(Howard et al., 1996; Puce et al., 1998) and vestibular perception
(Fasold et al., 2002; Friberg et al., 1985). Thus, it is possible that alpha
coherence reflects a coupling between the temporal cortex and
sensory regions of the basal ganglia nuclei (Brown et al., 1997). This
hypothesis is supported by a recent study showing that auditory
evoked fields can be modulated by stimulation of the STN (Airaksinen
et al., 2010). Interestingly, the hypothesis implies that frequency may
differentiate sensory from motor processing in STN-cortical loops.
Alpha coherence between STN and cortex may reflect sensory
processing in akinetic-rigid PD patients, whereas coherence in the
beta band reflects motor processing. However, this interpretation
remains speculative at this stage and requires further investigation.
Conclusions
By recording MEG and LFPs simultaneously we were able to
precisely map frequency-dependent interactions between STN and
cortex for the first time. Our study showed that STN-cortical
coherence is focal in the spatial and in the frequency domain and
revealed two distinct couplings between STN and cortex: One with
the motor cortex in the beta frequency band and one with temporal
areas in the alpha frequency band. Moreover, it further established
simultaneousMEG and intracranial electrode recordings as ameans to
study connectivity between deep and cortical brain areas in patients.
Acknowledgments
The authors would like to express their sincere gratefulness to the
patients who participated in this study. Furthermore, we are very
thankful to the people of Medtronic Neuromodulation (Dr. Ali Sarem-
Aslani, Mr. Paul van Venrooij, and Mr. Andreas Rolf) for technical
support. In addition, we thank Mrs. E. Rädisch for assistance with MRI
scans, Prof. Joachim Gross (CCNi Glasgow) for help in analysis issues
and the people behind the fieldtrip project for excellent support. This
study was supported by the ERANET-Neuron Grant “PhysiolDBS”
(Neuron-48-013) to A.S. The funding source did not influence the
collection, analysis and interpretation of data or the writing and
submission of the report. J.H. was supported by a travel grant from the
Böhringer Ingelheim Foundation (B.I.F.).
Appendix A. Supplementary data
Supplementary data to this article can be found online at
doi:10.1016/j.neuroimage.2010.11.063.
References
Ahonen, A., Hamalainen, M., Ilmoniemi, R., Kajola, M., Knuutila, J., Simola, J., Vilkman, V.,
1993. Sampling theory for neuromagnetic detector arrays. Biomed. Eng. IEEE Trans.
40, 859–869.Airaksinen, K., Mäkelä, J.P., Taulu, S., Ahonen, A., Nurminen, J., Schnitzler, A., Pekkonen,
E., 2010. Effects of DBS on auditory and somatosensory processing in Parkinson'sdisease. Hum. Brain Mapp. Electronic publication ahead of print.
Alexander, G.E., Crutcher, M.D., 1990. Functional architecture of basal ganglia circuits:
neural substrates of parallel processing. Trends Neurosci. 13, 266–271.Alexander, G.E., DeLong, M.R., Strick, P.L., 1986. Parallel organization of functionally
segregated circuits linking basal ganglia and cortex. Annu. Rev. Neurosci. 9,357–381.
Amjad, A.M., Halliday, D.M., Rosenberg, J.R., Conway, B.A., 1997. An extended differenceof coherence test for comparing and combining several independent coherence
estimates: theory and application to the study of motor units and physiological
tremor. J. Neurosci. Meth. 73, 69–79.Brillinger, D.R., 1981. Time series: data analysis and theory (SIAM).
Brown, L.L., Schneider, J.S., Lidsky, T.I., 1997. Sensory and cognitive functions of thebasal ganglia. Curr. Opin. Neurobiol. 7, 157–163.
Brown, P., Oliviero, A., Mazzone, P., Insola, A., Tonali, P., Di Lazzaro, V., 2001. Dopamine
dependency of oscillations between subthalamic nucleus and pallidum inParkinson's disease. J. Neurosci 21, 1033–1038.
Buchsbaum, B.R., Hickok, G., Humphries, C., 2001. Role of left posterior superiortemporal gyrus in phonological processing for speech perception and production.
Cogn. Sci. 25, 663–678.Cassidy, M., Mazzone, P., Oliviero, A., Insola, A., Tonali, P., Di Lazzaro, V., Brown, P., 2002.
Movement-related changes in synchronization in the human basal ganglia. Brain
125, 1235–1246.Djaldetti, R., Ziv, I., Melamed, E., 2006. The mystery of motor asymmetry in Parkinson's
disease. Lancet Neurol. 5, 796–802.Fasold, O., von Brevern, M., Kuhberg, M., Ploner, C.J., Villringer, A., Lempert, T., Wenzel,
R., 2002. Human vestibular cortex as identified with caloric stimulation in
functional magnetic resonance imaging. Neuroimage 17, 1384–1393.Fogelson, N., Williams, D., Tijssen, M., van Bruggen, G., Speelman, H., Brown, P., 2006.
Different functional loops between cerebral cortex and the subthalmic area inParkinson's disease. Cereb. Cortex 16, 64–75.
Friberg, L., Olsen, T.S., Roland, P.E., Paulson, O.B., Lassen, N.A., 1985. Focal increase ofblood flow in the cerebral cortex of man during vestibular stimulation. Brain 108
(Pt 3), 609–623.
Fries, P., 2005. A mechanism for cognitive dynamics: neuronal communication throughneuronal coherence. Trends Cogn. Sci. (Regul. Ed.) 9, 474–480.
Goetz, C.G., Tilley, B.C., Shaftman, S.R., Stebbins, G.T., Fahn, S., Martinez-Martin, P., Poewe,W., Sampaio, C., Stern, M.B., Dodel, R., et al., 2008. Movement Disorder Society-
sponsored revision of the Unified Parkinson's Disease Rating Scale (MDS-UPDRS): scale
presentation and clinimetric testing results. Mov. Disord. 23, 2129–2170.Gross, J., Kujala, J., Hamalainen, M., Timmermann, L., Schnitzler, A., Salmelin, R., 2001.
Dynamic imaging of coherent sources: studying neural interactions in the humanbrain. Proc. Natl. Acad. Sci. U. S. A. 98, 694–699.
Halliday, D.M., Rosenberg, J.R., Amjad, A.M., Breeze, P., Conway, B.A., Farmer, S.F., 1995.A framework for the analysis of mixed time series/point process data–theory and
application to the study of physiological tremor, single motor unit discharges and
electromyograms. Prog. Biophys. Mol. Biol. 64, 237–278.Hoover, J.E., Strick, 1999. The organization of cerebellar and basal ganglia outputs to
primary motor cortex as revealed by retrograde transneuronal transport of herpessimplex virus type 1. J. Neurosci. 19, 1446–1463.
Howard, Brammer, M., Wright, I., Woodruff, P.W., Bullmore, E.T., Zeki, S., 1996. A direct
demonstration of functional specialization within motion-related visual andauditory cortex of the human brain. Curr. Biol. 6, 1015–1019.
Howard, M., Volkov, I., Mirsky, R., Garell, P., Noh, M., Granner, M., Damasio, H.,Steinschneider, M., Reale, R., Hind, J., et al., 2000. Auditory cortex on the human
posterior superior temporal gyrus. J. Comp. Neurol. 416, 79–92.
Kühn, A.A., Williams, D., Kupsch, A., Limousin, P., Hariz, M., Schneider, G., Yarrow, K.,Brown, P., 2004. Event-related beta desynchronization in human subthalamic
nucleus correlates with motor performance. Brain 127, 735–746.Kühn, A.A., Trottenberg, T., Kivi, A., Kupsch, A., Schneider, G., Brown, P., 2005. The
relationship between local field potential and neuronal discharge in thesubthalamic nucleus of patients with Parkinson's disease. Exp. Neurol. 194,
212–220.
Kühn, A.A., Kempf, F., Brücke, C., Gaynor Doyle, L., Martinez-Torres, I., Pogosyan, A.,Trottenberg, T., Kupsch, A., Schneider, G., Hariz, M.I., et al., 2008. High-frequency
stimulation of the subthalamic nucleus suppresses oscillatory beta activity inpatients with Parkinson's disease in parallel with improvement in motor
performance. J. Neurosci. 28, 6165–6173.
1167J. Hirschmann et al. / NeuroImage 55 (2011) 1159–1168
Lalo, E., Thobois, S., Sharott, A., Polo, G., Mertens, P., Pogosyan, A., Brown, P., 2008.Patterns of bidirectional communication between cortex and basal ganglia duringmovement in patients with Parkinson disease. J. Neurosci. 28, 3008–3016.
Levy, R., Ashby, P., Hutchison, W.D., Lang, A.E., Lozano, A.M., Dostrovsky, J.O., 2002.Dependence of subthalamic nucleus oscillations on movement and dopamine inParkinson's disease. Brain 125, 1196–1209.
Litvak, V., Eusebio, A., Jha, A., Oostenveld, R., Barnes, G.R., Penny, W.D., Zrinzo, L., Hariz,M.I., Limousin, P., Friston, K.J., et al., 2010. Optimized beamforming forsimultaneous MEG and intracranial local field potential recordings in deep brainstimulation patients. Neuroimage 50, 1578–1588.
Marceglia, S., Foffani, G., Bianchi, A.M., Baselli, G., Tamma, F., Egidi, M., Priori, A., 2006.Dopamine-dependent non-linear correlation between subthalamic rhythms inParkinson's disease. J. Physiol. Lond. 571, 579–591.
Maris, E., Schoffelen, J., Fries, P., 2007. Nonparametric statistical testing of coherencedifferences. J. Neurosci. Meth. 163, 161–175.
Marsden, J.F., Limousin-Dowsey, P., Ashby, P., Pollak, P., Brown, P., 2001. Subthalamicnucleus, sensorimotor cortex and muscle interrelationships in Parkinson's disease.Brain 124, 378–388.
Mattout, J., Henson, R.N., Friston, K.J., 2007. Canonical source reconstruction for MEG.Comput. Intell. Neurosci. 67613.
Middleton, F.A., Strick, P.L., 1996. The temporal lobe is a target of output from the basalganglia. Proc. Natl. Acad. Sci. U. S. A. 93, 8683–8687.
Miyachi, S., Lu, X., Imanishi, M., Sawada, K., Nambu, A., Takada, M., 2006.Somatotopically arranged inputs from putamen and subthalamic nucleus toprimary motor cortex. Neurosci. Res. 56, 300–308.
Nambu, A., Takada, M., Inase, M., Tokuno, H., 1996. Dual somatotopical representationsin the primate subthalamic nucleus: evidence for ordered but reversed body-maptransformations from the primarymotor cortex and the supplementary motor area.J. Neurosci. 16, 2671–2683.
Nenonen, J., Taulu, S., Kajola, M., Ahonen, A., 2007. Total information extracted fromMEG measurements. Int. Congr. Ser. 1300, 245–248.
Nichols, T.E., Holmes, A.P., 2002. Nonparametric permutation tests for functionalneuroimaging: a primer with examples. Hum. Brain Mapp. 15, 1–25.
Nolte, G., 2003. The magnetic lead field theorem in the quasi-static approximation andits use for magnetoencephalography forward calculation in realistic volumeconductors. Phys. Med. Biol. 48, 3637–3652.
Pollok, B., Makhloufi, H., Butz, M., Gross, J., Timmermann, L., Wojtecki, L., Schnitzler, A.,2009. Levodopa affects functional brain networks in Parkinsonian resting tremor.Mov. Disord. 24, 91–98.
Priori, A., Foffani, G., Pesenti, A., Tamma, F., Bianchi, A.M., Pellegrini, M., Locatelli, M.,Moxon, K.A., Villani, R.M., 2004. Rhythm-specific pharmacological modulation ofsubthalamic activity in Parkinson's disease. Exp. Neurol. 189, 369–379.
Puce, A., Allison, T., Bentin, S., Gore, J.C., McCarthy, G., 1998. Temporal cortex activationin humans viewing eye and mouth movements. J. Neurosci. 18, 2188–2199.
Rodriguez-Oroz, M.C., Rodriguez, M., Guridi, J., Mewes, K., Chockkman, V., Vitek, J.,DeLong, M.R., Obeso, J.A., 2001. The subthalamic nucleus in Parkinson's disease:somatotopic organization and physiological characteristics. Brain 124, 1777–1790.
Rosa, M., Marceglia, S., Servello, D., Foffani, G., Rossi, L., Sassi, M., Mrakic-Sposta, S.,Zangaglia, R., Pacchetti, C., Porta,M., et al., 2010. Timedependent subthalamic localfieldpotential changes after DBS surgery in Parkinson's disease. Exp. Neurol. 222, 184–190.
Schaltenbrand, G., Wahren, W., 1977. Atlas for stereotaxy of the human brain withguide to the atlas for stereotaxy of the human brain, Pkg Lslf. (Thieme).
Schnitzler, A., Gross, J., 2005. Normal and pathological oscillatory communication in thebrain. Nat. Rev. Neurosci. 6, 285–296.
Schoffelen, J., Gross, 2009. Source connectivity analysis with MEG and EEG. Hum. BrainMapp. 30, 1857–1865.
Schoffelen, J., Oostenveld, R., Fries, P., 2008. Imaging the human motor system's beta-band synchronization during isometric contraction. Neuroimage 41, 437–447.
Selemon, L., Goldman-Rakic, P., 1985. Longitudinal topography and interdigitation ofcorticostriatal projections in the rhesus monkey. J. Neurosci. 5, 776–794.
Song, T., Gaa, K., Cui, L., Feffer, L., Lee, R.R., Huang, M., 2008. Evaluation of signal spaceseparation via simulation. Med. Biol. Eng. Comput. 46, 923–932.
Srinivasan, R., Winter, W.R., Ding, J., Nunez, P.L., 2007. EEG and MEG coherence:measures of functional connectivity at distinct spatial scales of neocorticaldynamics. J. Neurosci. Meth. 166, 41–52.
Taulu, S., Kajola, M., 2005. Presentation of electromagnetic multichannel data: thesignal space separation method. J. Appl. Phys. 97, 124905.
Taulu, S., Simola, J., Kajola, M., 2005. Applications of the signal space separationmethod.Signal Process. IEEE Trans. 53, 3359–3372.
Timmermann, L., Gross, J., Dirks, M., Volkmann, J., Freund, H., Schnitzler, A., 2003. Thecerebral oscillatory network of parkinsonian resting tremor. Brain 126, 199–212.
Trottenberg, T., Kupsch, A., Schneider, G., Brown, P., Kühn, A.A., 2007. Frequency-dependent distribution of local field potential activity within the subthalamicnucleus in Parkinson's disease. Exp. Neurol. 205, 287–291.
Van Hoesen, G.W., Yeterian, E.H., Lavizzo-Mourey, R., 1981. Widespread corticostriateprojections from temporal cortex of the rhesus monkey. J. Comp. Neurol. 199,
205–219.
Van Veen, B.D., van Drongelen, W., Yuchtman, M., Suzuki, A., 1997. Localization of brainelectrical activity via linearly constrained minimum variance spatial filtering. IEEE
Trans. Biomed. Eng. 44, 867–880.Weinberger, M., Mahant, N., Hutchison, W.D., Lozano, A.M., Moro, E., Hodaie, M., Lang,
A.E., Dostrovsky, J.O., 2006. Beta oscillatory activity in the subthalamic nucleus andits relation to dopaminergic response in Parkinson's disease. J. Neurophysiol. 96,
3248–3256.
Williams, D., Tijssen, M., Van Bruggen, G., Bosch, A., Insola, A., Di Lazzaro, V., Mazzone,P., Oliviero, A., Quartarone, A., Speelman, H., et al., 2002. Dopamine-dependent
changes in the functional connectivity between basal ganglia and cerebral cortex inhumans. Brain 125, 1558–1569.
Winter, W.R., Nunez, P.L., Ding, J., Srinivasan, R., 2007. Comparison of the effect of
volume conduction on EEG coherence with the effect of field spread on MEGcoherence. Stat. Med. 26, 3946–3957.
Yeterian, E.H., Van Hoesen, G.W., 1978. Cortico-striate projections in the rhesusmonkey: the organization of certain cortico-caudate connections. Brain Res. 139,
43–63.
1168 J. Hirschmann et al. / NeuroImage 55 (2011) 1159–1168
J. Hirschmann et al. / NeuroImage 55 (2011) 1159–1168 supplementary. - page 1
Supplementary Material
Table S1 lists the stereotactic coordinates of all contacts from all patients.
Figure S2 demonstrates that each subject showed distinct alpha and beta peaks in MEG-LFP
coherence on the sensor level.
J. Hirschmann et al. / NeuroImage 55 (2011) 1159–1168 supplementary. - page 2
S1: Stereotactic coordinates of all DBS electrode contacts mm referenced to mid-commisural point
(Schaltenbrand-Wahren coordinate system).
x y z x y z
left 0 -11.1 -3.5 -2.2 -13.6 -3.3 -1.7
left 1 -11.9 -1.7 -0.4 -14.2 -1.7 0.0
left 2 -12.6 0.0 1.4 -14.9 -0.1 1.7
left 3 -13.4 1.8 3.2 -15.6 1.6 3.4
right 0 9.3 -5.2 -3.7 13.8 -1.7 -1.7
right 1 10.1 -3.4 -1.9 14.5 -0.1 0.0
right 2 10.8 -1.7 -0.1 15.2 1.5 1.7
right 3 11.6 0.1 1.7 15.9 3.2 3.4
Patient BH HH
x y z x y z
left 0 -12.5 -2.6 -2.9 -12.2 0.4 -3.9
left 1 -13.1 -1.1 -1.4 -12.9 2.0 -2.3
left 2 -13.7 0.4 0.2 -13.5 3.5 -0.7
left 3 -14.4 1.9 1.7 -14.2 5.1 0.9
right 0 13.2 -4.3 -2.4 11.4 1.4 -5.6
right 1 13.8 -2.8 -0.9 12.1 3.0 -4.0
right 2 14.4 -1.3 0.7 12.7 4.5 -2.4
right 3 15.1 0.2 2.2 13.4 6.1 -0.8
Patient KH SA
x y z x y z
left 0 -8.5 -5.1 -2.6 -12.5 -4.3 -4.2
left 1 -9.1 -3.7 -1.1 -13.1 -2.8 -2.6
left 2 -9.7 -2.3 0.3 -13.8 -1.3 -1.1
left 3 -10.3 -0.9 1.8 -14.4 0.2 0.5
right 0 10.9 -4.7 -3.0 10.8 -5.0 -4.8
right 1 11.5 -3.3 -1.5 11.4 -3.5 -3.2
right 2 12.1 -1.9 -0.1 12.1 -2.0 -1.7
right 3 12.7 -0.5 1.4 12.7 -0.5 -0.1
Patient ZE LW
x y z x y z
left 0 -10.3 -4.1 -2.6 -10.5 -1.3 -3.7
left 1 -11.0 -2.5 -0.9 -11.2 0.4 -2.0
left 2 -11.7 -0.9 0.7 -11.9 2.1 -0.2
left 3 -12.4 0.7 2.4 -12.7 3.8 1.5
right 0 12.0 -4.1 -2.5 13.2 -3.4 -5.4
right 1 12.7 -2.5 -0.8 13.9 -1.7 -3.7
right 2 13.4 -0.9 0.8 14.6 0.0 -1.9
right 3 14.1 0.7 2.5 15.4 1.7 -0.2
Patient PH WA
x y z x y z
left 0 -11.4 -3.0 -3.0 1.61 1.78 0.88
left 1 -12.1 -1.4 -1.3 1.62 1.82 0.92
left 2 -12.7 0.2 0.3 1.63 1.86 0.97
left 3 -13.4 1.8 1.9 1.64 1.91 1.03
right 0 11.8 -3.4 -3.6 1.52 2.22 1.48
right 1 12.5 -1.8 -2.0 1.51 2.23 1.47
right 2 13.2 -0.2 -0.4 1.51 2.24 1.48
right 3 13.8 1.4 1.3 1.51 2.26 1.49
Patient average standard deviation
J. Hirschmann et al. / NeuroImage 55 (2011) 1159–1168 supplementary. - page 3
S2: Examples of MEG-LFP coherence on the sensor level. For each subject, the two coherence spectra
with the most prominent peaks in the alpha and beta band, respectively, are depicted. The positions of
the corresponding MEG sensors are indicated in the topographical plots of the MEG sensor array.
Black squares relate to the spectra showing alpha peaks, red circles relate to the spectra showing beta
peaks. The corresponding LFP channels are given above the coherence spectra.
References
Schaltenbrand, G., and Wahren, W. (1977). Atlas For Stereotaxy of The Human Brain With
Guide to The Atlas For Stereotaxy of The Human Brain, Pkg Lslf. (Thieme).
Differential modulation of STN-cortical and cortico-muscular coherence by movementand levodopa in Parkinson's disease
J. Hirschmann a,b, T.E. Özkurt c, M. Butz a,d, M. Homburger a,b, S. Elben a,b, C.J. Hartmann a,b, J. Vesper e,L. Wojtecki a,b, A. Schnitzler a,b,⁎a Institute of Clinical Neuroscience and Medical Psychology, Medical Faculty, Heinrich-Heine-University Düsseldorf, Düsseldorf, Germanyb Department of Neurology, University Hospital Düsseldorf, Düsseldorf, Germanyc Department of Health Informatics, Informatics Institute, Middle East Technical University, Ankara, Turkeyd Sobell Department of Motor Neuroscience and Movement Disorders, Institute of Neurology, University College London, London, UKe Department of Functional Neurosurgery and Stereotaxy, University Hospital Düsseldorf, Düsseldorf, Germany
a b s t r a c ta r t i c l e i n f o
Article history:
Accepted 15 November 2012Available online 16 December 2012
Keywords:
Parkinson's diseaseOscillationsMEGCoherenceSTNDeep brain stimulation
Previous research suggests that oscillatory coupling between cortex, basal ganglia and muscles plays an im-portant role in motor behavior. Furthermore, there is evidence that oscillatory coupling is altered in patientswith movement disorders such as Parkinson's disease (PD).In this study, we performed simultaneous magnetoencephalography (MEG), local field potential (LFP) andelectromyogram (EMG) recordings in PD patients selected for therapeutic subthalamic nucleus (STN) stimu-lation. Patients were recorded (i) after withdrawal of anti-parkinsonian medication (OFF) and (ii) after levo-dopa administration (ON). We analyzed STN-cortical and cortico-muscular coherence during static forearmcontraction and repetitive hand movement in order to evaluate modulations of coherence by movementand medication. Based on previous results from studies investigating resting state coherence in PD patients,we selected primary motor cortex (M1) and superior temporal gyrus (STG) as regions of interest.We found beta coherence between M1 and STN to be suppressed by administration of levodopa. M1–muscularcoherencewas strongly reduced in the alpha and beta band during repetitivemovement compared to static con-traction, but was unaffected by administration of levodopa. Strong STG–STN but not STG–muscular coherencecould be observed in the alpha band. Coherence with STG was modulated neither by movement nor by medica-tion. Finally, we found both M1–STN and M1–muscular beta coherence to be negatively correlated with UPDRSakinesia and rigidity sub-scores in the OFF state.The present study provides new insights into the functional roles of STN-cortical and cortico-muscular coherenceand their relationship to PD symptoms. The results indicate that STN-cortical and cortico-muscular coupling arecorrelated, but can bemodulated independently. Moreover, they showdifferences in their frequency-specific to-pography. We conclude that they represent partly independent sub-loops within the motor system. Given theirnegative correlation with akinesia, neither can be considered “antikinetic”.
© 2012 Elsevier Inc. All rights reserved.
Introduction
Parkinson's disease (PD) is a neurodegenerative disorder resultingfroma loss of dopaminergic neuronswhich leads to alterations of neuralactivity in the basal ganglia, thalamus and cortex (Lang and Lozano,1998). Recent research pinpointed excessive synchronization as anelectrophysiological hallmark of PD (for a review see Hammond et al.,2007). Local field potential (LFP) recordings from the subthalamic nu-cleus (STN) of patients undergoing surgery for deep brain stimulation(DBS) revealed strong oscillatory activity, particularly in the beta band
(13-35 Hz). STN beta oscillations were found to be reduced by applica-tion of levodopa (Brown et al., 2001; Kühn et al., 2006; Levy et al., 2002;Priori et al., 2004), movement (Cassidy et al., 2002; Kühn et al., 2004)and DBS (Eusebio et al., 2011; Giannicola et al., 2010; Kühn et al.,2008). Furthermore, it was shown that STN beta power reduction corre-lates with clinical improvement (Kühn et al., 2008). Given their promi-nence in static states, beta oscillations have been labeled “antikinetic”.In turn, gamma oscillations (60–90 Hz) have been labeled “prokinetic”and are considered the functional counterpart of beta oscillations(Brown, 2003). They increase in power with both anti-parkinsonianmedication and movement (Cassidy et al., 2002; Williams et al., 2002).
A similar pattern of responses is observable in other brain areaswhich are part of the motor network and in oscillatory coupling be-tween areas. For example, movement reduces beta band coherencebetween STN and sensorimotor cortex (Lalo et al., 2008; Marsden
NeuroImage 68 (2013) 203–213
⁎ Corresponding author at: Institute of Clinical Neuroscience and Medical Psycholo-gy, Heinrich-Heine-University Düsseldorf, Universitätsstr. 1, D-40225 Düsseldorf, Ger-many. Fax: +49 211 811 9033.
E-mail address: [email protected] (A. Schnitzler).
1053-8119/$ – see front matter © 2012 Elsevier Inc. All rights reserved.http://dx.doi.org/10.1016/j.neuroimage.2012.11.036
Contents lists available at SciVerse ScienceDirect
NeuroImage
j ourna l homepage: www.e lsev ie r .com/ locate /yn img
et al., 2001b), the latter of which was reported to show enhancedbeta power in PD (Pollok et al., 2012). Likewise, beta coherence be-tween sensorimotor cortex and muscle activity is suppressed duringmovement (Baker et al., 1997; Kilner et al., 1999). The questionarises whether these observations reflect network responses, affect-ing all couplings between motor areas alike, or whether oscillatorycoupling may be modulated independently and specifically for anytwo elements of the network.
An example for independent modulations may be given by the re-sponses of STN-cortical and cortico-muscular beta coherence to PD treat-ment. Cortico-muscular beta coherence was reported to be enhancedby levodopa (Salenius et al., 2002) and DBS was found to increaseintermuscular beta coherence (Marsden et al., 2001a). STN-corticalbeta coherence, on the other hand, was reported to be decreasedby administration of levodopa (Sharott et al., 2005; Williams et al.,2002) and DBS (Kühn et al., 2008).
However, recent studies could not confirm the effect of pharmaco-logical intervention on STN-cortical coherence. Lalo et al. (2008) in-vestigated the direct transfer function between STN and cortex anddid not detect a difference in beta band interactions after levodopawas administered. Along the same lines, Litvak et al. (2011) couldnot detect a decrease in beta coherence either, but reported an in-crease in a small area in prefrontal cortex.
In summary, the role of STN-cortical coupling in the pathophysiolo-gy of PD remains to be elucidated. Currently, it is unclear if and how it ismodulated by levodopa, how such modulations relate to clinical symp-toms and whether it changes independently from cortico-muscular co-herence. In this study, we aimed to answer these questions by analysisof simultaneousmagnetoencephalography (MEG), LFP and electromyo-gram (EMG) recordings. We studied STN-cortical and cortico-muscularcoherence during epochs of repetitivemovement and static contractionbefore and after the administration of levodopa. Our analyses provide acomprehensive and comparative description of STN-cortical and cortico-muscular coherence in PD patients and give new insights into their rela-tion to clinical symptoms.
Materials and methods
Patients
10 PD patients (four female), who were clinically selected for deepbrain stimulation (DBS) because of levodopa-induced fluctuations anddyskinesias, participated in this study with written informed consent.For all patients, the dominant symptoms were akinesia and rigidity.MEG and LFP data from eight of these subjects (rest condition, off med-ication) were analyzed in a previous study (Hirschmann et al., 2011).Table 1 summarizes the clinical details. The study was approved bythe local ethics committee (study no. 3209) and is in accordance withthe Declaration of Helsinki.
Surgery
The implantation of macroelectrodes was carried out at the De-partment of Functional Neurosurgery and Stereotaxy of the Univer-sity Hospital Düsseldorf. The surgical procedures are described elsewhere(Özkurt et al., 2011). Oral anti-parkinsonian medication was withdrawnthe evening before surgery and substituted by subcutaneous apomor-phine medication. All but one subject were implanted with electrodemodel 3389 (Medtronic Inc., Minneapolis, MN, USA). Subject 9 wasimplanted with a DBS system by St. Jude Medical (St. Jude Medical Inc.,St. Paul, MN, USA). Electrode placement was guided by intraoperativemicroelectrode recordings, intraoperative stimulation and clinical test-ing. Correct placementwas confirmed by stereotactic, postsurgical com-puter tomography (CT). Fig. S1 of the supplementary material shows aprojection of the average, normalized coordinates of the contacts used
in this study onto the Schaltenbrand–Wahren atlas (Schaltenbrand andWahren, 1977).
Recordings
We simultaneously recorded MEG, LFPs from the STN, EMG of theextensor digitorum communis and flexor digitorum superficialismusclesof both upper limbs and vertical and horizontal electrooculograms(EOGs). All recordings were performed using a 306 channel, whole-head MEG system (Elekta Oy, Helsinki, Finland). The sampling ratewas 2 kHz. For all but one subject (subject 10) non-magnetic exten-sion leads were available to connect the DBS macroelectrodes to theEEG amplifier integrated into the MEG system. MEG signals were fil-tered online between 0.03 and 660 Hz, while LFP, EMG and EOG sig-nals were filtered between 0.1 and 660 Hz. EMGs were referenced tosurface electrodes at the muscle tendons. LFPs were referenced to asurface electrode at the left mastoid and rearranged into a bipolarmontage offline. This was done by subtracting from each of the fourmonopolar signals the signal of its ventral neighbor channel, so thatthree bipolar channels per electrode were obtained
Paradigm
Recordings took place the day after surgery. Two hours before re-cording the apomorphine pump was switched off. The experimenthad a block design. It included (i) one block off dopaminergic medica-tion (OFF) and (ii) one block following the administration of levodopa(ON). The clinical OFF state was quantified by means of the motorscore of the Movement Disorder Society Unified Parkinson's DiseaseRating Scale (MDS-UPDRS III) immediately before the recordingstarted (Goetz et al., 2008). The rating was performed by an experi-enced movement disorders specialist and recorded with video cam-era for an additional offline rating by a second rater.
The experiment startedwith a 5 min rest recording (REST), followedby two motor tasks. Subjects had their eyes open. In the first task(HOLD), subjects were asked to elevate one forearm to about 45° withthe elbow resting on a table in front of them and to spread their fingers.In the second task (MOVE), subjects were instructed to open and closeone fist. The arm was elevated to about 45°, as in the HOLD condition.Subjects were asked to perform self-paced movements with a frequen-cy of approximately 1 Hz. Task performance was monitored online bythe experimenters via a video camera and offline by visual investigationof EMG signals. In all tasks solely the symptom-dominant hand wasused. Each motor task lasted 9 min: Five times 1 min of task perfor-mance interleaved by four pauses of 1 min duration. Pauses served toavoid muscle fatigue.
After completion of all three aforementioned conditions in the OFFstate, levodopa was administered (Madopar LT, Roche Pharma AG,Basel, Switzerland). The doses are listed in Table 1. They were chosento be approximately 1.5 times the pre-surgical morning dose of eachindividual subject to obtain a good clinical ON state. In several casesdoses were adjusted by the neurologist supervising the experimentin order to minimize levodopa-induced dyskinesias, which occurredin five subjects. After waiting for ≥30 min, we tested upper limb ri-gidity and akinesia. In case an improvement was observable, anotherMDS UPDRS motor score was performed and videotaped to quantifythe subjects' ON state and the motor tasks were repeated.
Apart from the experimenters' instructions and corrections, no vi-sual or auditory stimuli were presented.
Data preprocessing
Data were notch filtered at 50 Hz to remove power line noise.EMGs were high-pass filtered at 10 Hz and full-wave rectified (Kilneret al., 2000). All data were down-sampled to 256 Hz and divided intohalf-overlapping segments of 256 samples. The 5 min block of motor
204 J. Hirschmann et al. / NeuroImage 68 (2013) 203–213
task execution, which was interleaved by 1 min pauses, was treatedas one continuous recording. Signals underwent thorough visual ex-amination. Artifacts and bad channels were excluded from analysis.Reasons for data exclusion were postural tremor (3 subjects), diffi-culties in performing the required movement (2 subjects) and con-tamination by noise from magnetic extension leads (1 subject; seebelow). Preprocessing and artifact removal yielded epochs of clean datawith a mean duration of 257 s per subject and motor task (range: 137–330 s). EMGs were used to identify periods of holding and moving ofthe hand and pauses between epochs of task execution were discarded.
Subject 10 was recorded with magnetic extension leads. MEG datafor this subjectwere contaminated by artifacts. Thus, we applied tempo-
ral Signal Space Separation (tSSS) specifically on this dataset in order tosuppress the magnetic artifacts (Taulu and Simola, 2006). Previously,tSSS has been used successfully to suppressmagnetic artifacts occurringduring combined MEG and DBS (Airaksinen et al., 2011, 2012). Thelimits of spherical harmonic expansion were chosen as Lin=8 for theinner and Lout=3 for the outer sources, as these values have beenshown to be optimal in an earlier study (Taulu et al., 2005). Visual ex-amination confirmed that tSSS resulted in acceptable data quality, ex-cept for some epochs which were discarded (mean duration of cleandata per condition: 205 s).
Data analysis
Only the 204 gradiometer channels were used for the analysis ofMEG signals. Data analysis was performed with Matlab R2010 (The
Mathworks, Natick, Massachusetts, USA) and we made use of theFieldTrip toolbox (Oostenveld et al., 2011). ANOVAs were computedusing the software package IBM SPSS Statistics 19 (IBM Corporation,Somers, USA).
Regions of interest and channel selection
We estimated coherence between two cortical regions of interest(ROIs) and an LFP and EMG reference channel, respectively. In ordernot to bias statistical analysis, we did not choose ROIs and referencechannels based on maxima observed in any of the two motor tasks.Instead, we used maxima occurring in the rest recordings precedingthe motor tasks and excluded these recordings from subsequent analy-sis, i.e. rest recordings served as independent localizer sessions. Thus,we avoided the risk of selecting reference channels and ROIs in whichhigh values occurred by chance in only one of several conditions(Saxe et al., 2006).
The spatial maxima of STN-cortical coherence at rest have been stud-ied previously in a subset of the subjects included here (Hirschmannet al., 2011). In addition, resting state coherence in PD has been investi-gated in an independent study (Litvak et al., 2011). Both studies reportedstrong beta coherence between STN and sensorimotor areas, includingprimary motor cortex (M1). Alpha coherence was found predominantlywith areas in the temporal lobe, in particular with superior temporalgyrus (STG). Accordingly, we chose M1 and STG as ROIs. The MNI coor-dinates were±33, -22, 57 for M1 and±51, -20, 5 for STG (Hirschmann
et al., 2011). Cortico-muscular coherence was measured at the same
Table 1
Clinical details of patients.
Stereotactic coordinates are given in millimeters and are defined according to the Schaltenbrand–Wahren coordinate system. Since recordings were bipolar, the table lists the av-
erage coordinates of two neighboring contacts for each subject. n.a. = not available.
Subject Age Sex Disease
duration
(years)
UPDRS OFF
presurgical
UPDRS OFF
postsurgical
UPDRS ON
postsurgical
Pre-surgical medication
(daily dose in mg)
Administered dose of
levodpa (mg)
Coordinates of contact for LFP
recordings (X Y Z)
1 68 M 11 34 43 29 Levodopa 400
Entacapone 800
Pramipexole 1.4
150 −14.0 −0.8 0.8
2 75 F 26 n.a. 46 19 Levodopa 800
Ropinirole 18
Rasagiline 1
Tolcapone 100
150 8.6 −3.8 −2.5
3 71 M 16 45 48 34 Levodopa 1000
Entacapone 1200
Amantadine 100
Ropinirole 10
200 13.1 5.3 −1.6
4 70 M 11 34 46 17 Levodopa 825
Entacapone 1000
Pramipexole 3.15
Rasagiline 1
250 −14.1 −2.1 −1.9
5 61 M 16 39 43 27 Levodopa 825
Entacapone 1000
Selegiline 10
200 −13.3 −1.9 −2.2
6 62 M 21 39 61 33 Levodopa 950
Entacapone 1000
Ropinirol 24
Rasagiline 1
250 −11.1 −1.8 1.1
7 64 F 18 38 30 11 Levodopa 900
Amantadine 400
Pramipexole 1.4
Rasagiline 1
150 13.3 −0.8 −2.6
8 62 F 15 44 27 22 Levodopa 450
Pramipexole 0.54
150 13.3 −0.1 1.6
9 53 M 11 n.a. 30 15 Levodopa 300
Amantadine 100
300 −12.0 −0.6 −4.9
10 55 F 10 44 46 13 Levodopa 800
Tolcapone 300
Rasagiline 1
Ropinirole 24
200 −13.1 −0.2 0.8
Mean 64.1 15.5 40 42 22 200 12.6 −0.7 −1.1
Std 7.0 5.2 4.4 10.3 8.3 52.7 1.7 2.4 2.1
205J. Hirschmann et al. / NeuroImage 68 (2013) 203–213
location so that it could be directly compared to STN-cortical coherence.
Themacroelectrode contact pair yielding highest beta coherence with a
group of sensorimotor MEG channels at rest was chosen as the refer-
ence channel for STN-cortical coherence computations. Only contact
pairs contralateral to the moved limb were considered. As reference
channel for cortico-muscular coherence,we chose the EMGof the exten-
sor digitorum communismuscle of the moved limb.
Spectral analysis
Estimation of spectral parameters was performed using the
multitapering approach (Thomson, 1982). We studied three frequency
bands: alpha (8–12 Hz), beta (13–35 Hz) and gamma (60–90 Hz). In
order to obtain a single value for an entire band, we computed spectral
estimates for the central frequency of each bandwith appropriate spec-
tral smoothing to cover the entire band. For computation of complete
spectra, we convolved data with a Hanning taper and applied Welch's
method.
Source reconstruction
Estimation of power and coherence on the source level was realized
by Dynamic Imaging of Coherent Sources (DICS; Gross et al., 2001), a fre-
quency domain beamformer. Orientation of the reconstructed activity
was defined as the orientation that maximized power. Regularization
was not applied. DICS projects sensor data through a spatial filter de-
rived from the forwardmodel and theMEG sensor cross spectral densi-
ties. We used all artifact-free gradiometers for filter construction. The
forward model was based on a realistic, single shell head model
derived from individual T1-weighted structural magnetic reso-
nance images (Nolte, 2003). The latter were obtained prior to sur-
gery using aMagnetomTrioMRI scanner (Siemens, Erlangen,Germany)
and 3D magnetization-prepared rapid gradient-echo imaging (repeti-
tion time: 2300 ms, echo time: 2.98 ms).
Analysis of effects of medication and motor task
Effects of medication and motor task were investigated by repeated
measures analysis of variance (ANOVA). Coherence values were Fisher
z-transformed prior to computing ANOVAs since the distribution of
the transform is closer to normal than that of non-transformed coher-
ence (Halliday et al., 1995). The number of data segments considered
in statistical analysis was intra-individually balanced across experi-
mental conditions by choosing an equally long interval from all con-
ditions. A three-way ANOVA was performed for each frequency band.
The factors were shuffling (ORIGINAL, SHUFFLED), medication (ON,
OFF) and motor task (HOLD, MOVE). In the SHUFFLED condition, the
signal of the reference channel was shifted forward in time by k seg-
ments (circular shift). k was a random integer between 2 and N−1, N
being the total number of segments. True physiological effects are
expected to interact with shuffling as it destroys condition specific
coherence differences present in the original data. In case an effect
interacted with shuffling, we performed two-way, follow-up ANOVAS
for the ORIGINAL and the SHUFFLED condition separately.
Correlation between coherence and clinical symptoms
In order to investigate the relation between coherence and clinical
symptoms we computed Pearson's linear correlation coefficient be-
tween Fisher z-transformed coherence values and the sum of the
hemibody akinesia/rigidity UPDRS sub-scores (part 3, items 3.3b–
3.8b, body side specific for the moved limb). We used the average of
both ratings, i.e. the online and the offline UPDRS rating for correlation.
To assure that theobserved correlationswere independent of clinical im-
provements induced by electrode implantation (stun effect) we repeat-
ed correlations using pre-surgical UPDRS scores. These were obtained
five days prior to surgery by a movement disorders specialist and were
also re-rated offline. Recent pre-surgical OFF scores were unavailable
for two cases (subjects 2 and 9).
Statistical tests of correlation were corrected for multiple compar-
isons by the Benjamini–Hochberg procedure for false discovery rate
control. All p-values were adjusted.
Results
There was a main effect of shuffling on coherence for all signal pairs
and frequency bands investigated (F≥6.92, p≤0.03), except forM1-STN
gamma coherence (F=2.42, p=0.16) and STG–muscular alpha coher-
ence (F=3.73, p=0.09). In the following, we will report all significant
interactions with shuffling and the results of the corresponding follow-
up ANOVAs. Table 2 summarizes the results of the follow-up ANOVAs
for the ORIGINAL condition. The degrees of freedom were 1 (between
measures) and 9 (error).
Coherence between M1 and STN
We found an interaction between shuffling and medication
for M1-STN beta coherence (F=11.27, pb0.01). The follow-up
ANOVAs revealed a significant effect of medication in the ORIGINAL
(F=11.66, pb0.01) but not in the SHUFFLED condition (F=1.62,
p=0.24). In the ORIGINAL condition, a statistical trend was observ-
able for an interaction between medication and motor task (F=4.08,
p=0.07). As shown in Fig. 1, administration of levodopa reduced
mean M1–STN beta coherence. This effect was more pronounced in
the HOLD than in the MOVE condition, in which beta coherence was
comparably weak.
Fig. 2 illustrates the spectral changes which occurred in response
to levodopa intervention in two individual subjects and the group av-
erage. The choice of displayed single subject spectra was based on
similarity to group data. The strength of the dopamine-induced re-
duction and the exact frequencies at which it occurred were variable
across subjects, so that the average spectra do not exhibit very clear
differences (Figs. 2B and D). Moreover, the effect of levodopa was
not limited to the beta band in all cases. Individual subjects showed
an additional decrease in alpha coherence or an increase in gamma
(60–90 Hz) coherence (Fig. 2A).
Coherence between M1 and muscle
In contrast to M1–STN coherence, M1–muscular coherence showed
strong modulation by movement rather than medication. Shuffling
interacted with motor task in the alpha (F=11.12, pb0.01) and
beta band (F=17.79, pb0.01). The effect of motor task was significant
in the ORIGINAL (alpha: F=8.41, p=0.02; beta: F=18.9, pb0.01)
but not in the SHUFFLED condition (alpha: F=4.45, p=0.06; beta:
Table 2
ANOVA tables for the original data.
Effects which showed an interaction with shuffling in the foregoing three-way ANOVA
are printed in italics.
Beta coherence M1–STN Alpha coherence M1–muscle
F p F P
Medication 11.66 b0.01 Medication 1.38 0.27
Motor task 2.89 0.12 Motor task 8.41 0.02
Medication∗motor task 4.08 0.07 Medication∗motor task 0.01 0.94
Gamma coherence STG–STN Beta coherence M1–muscle
F p F P
Medication 3.61 0.09 Medication 0.25 0.63
Motor task 1.40 0.27 Motor task 18.90 b0.01
Medication∗motor task 0.48 0.51 Medication∗motor task 0.04 0.86
206 J. Hirschmann et al. / NeuroImage 68 (2013) 203–213
F=0.05, p=0.83). As depicted in Fig. 1, mean cortico-muscular alpha
and beta coherencewas smaller in theMOVE than in theHOLD condition.
Fig. 3 illustrates the spectral changes induced by movement in two
individual subjects and the group average. The effect of motor task on
M1–muscular coherence was more broad-band and more consistent
across subjects than the effect of levodopa on M1–STN coherence
(Figs. 3B and D). It was equally strong in the OFF and the ON state.
Coherence between STG and STN
Fig. 4 depicts the average STG–STN and STG–muscular coherence
spectra. STG showed strong alpha band coherence with STN in all
experimental conditions, as evidenced by a strong main effect of
shuffling (F=14.56, pb0.01). STG–STN coherence was neither modu-
lated by administration of levodopa nor bymotor task in any frequency
band. We did find a significant interaction between shuffling and med-
ication in the gamma band (F=5.80, p=0.04). However, the effect of
medication was not significant in the ORIGINAL condition (F=3.61,
p=0.09).
Coherence between STG and muscle
In contrast to STG–STN coherence, shuffling did not affect alpha
coherence between STG and muscle (F=3.73, p=0.09), indicating
Fig. 1. Effects of levodopa and motor task on mean coherence with M1. Coherence values were z-transformed and averaged across subjects. Error bars indicate the standard error of
the mean (SEM). Asterisks mark significant differences (pb0.05). Please note the smaller scaling for the gamma band.
Fig. 2. Administration of levodopa reduces beta coherence between M1 and STN. A) Coherence spectra of subject 6 in the HOLD condition. Please note the gamma band increase
co-occurring with a beta band decrease. B) Average coherence in the HOLD condition. Shaded areas indicate SEMs. C) Coherence spectra of subject 2 in the MOVE condition.
D) Average coherence in the MOVE condition.
207J. Hirschmann et al. / NeuroImage 68 (2013) 203–213
that muscle activity was not coherent with STG activity in the alpha
band. While there was a main effect of shuffling when considering
beta or gamma coherence, no significant interactions with shuffling
were detected in any frequency band.
MEG, LFP and EMG power and STN–muscular coherence
We investigated whether modulations of power were similar to
modulations of coherence. To limit the amount of statistical tests,
only signals showing significant modulations of coherence in the pre-
vious analysis were considered. Fig. S2 of the supplementary material
shows the condition-specific means. M1 alpha and beta power were
neither affected by medication nor by motor task. STN beta power
was reduced by administration of levodopa (F=12.65, pb0.01) and
showed a trend for reduction by movement (F=4.6, p=0.06). EMG
alpha (F=6.96. p=0.03) and beta (F=7.09. p=0.03) power were
increased during MOVE compared to HOLD.
In addition, we investigated modulations of coherence between
STN andmuscle. STN–muscular coherence was reduced bymovement
in the beta band (F=11.9, pb0.01) but was not affected by adminis-
tration of levodopa (Fig. S3).
Correlations with UPDRS
The reduction of M1–STN beta coherence induced by levodopa did
not correlate with the reduction of akinesia/rigidity hemibody UPDRS
Fig. 3.Movement reduces alpha and beta coherence between M1 and muscle. A) Coherence spectra of subject 10 in medication OFF. B) Average coherence in medication OFF. Shad-
ed areas indicate SEMs. C) Coherence spectra of subject 7 in medication ON. D) Average coherence in medication ON.
Fig. 4. STN but not muscle activity is coherent with STG. Average coherence between STG and STN (left) and STG and muscle (right). Shaded areas indicate SEMs.
208 J. Hirschmann et al. / NeuroImage 68 (2013) 203–213
score, nor did coherence and UPDRS correlate in the ON state (|r|≤0.29,
p≥0.52). In the OFF state, however, M1–STN beta coherence correlated
negatively with UPDRS score, i.e. the subjects with the least
akinesia showed the strongest coupling. The anti-correlation was
significant in the MOVE OFF condition (r=−0.82, p=0.04) and
a trend was observable in the HOLD OFF condition (r=−0.67,
p=0.08). For coherence between M1 and muscle, a similar pat-
tern was observed (rHOLD_OFF=−0.70, p=0.07; rMOVE_OFF=−0.73,
p=0.06; Fig. 5). The correlation was specific to the beta band. Neither
alpha (rM1_STN=0.01, p=0.99; rM1_MUSCLE=−0.48, p=0.25) nor
gamma band coherence (rM1_STN=−0.34, p=0.46; rM1_MUSCLE=
0.00, p=0.99) showed a similar relation to UPDRS scores (Fig. 6).
Furthermore, the correlation could not be attributed to changes induced
by electrode implantation. Like post-surgical scores, pre-surgical OFF
scores were negatively correlated with M1–STN beta coherence
(r=−0.88, p=0.04; Fig. S4).
The similarity between M1–STN and M1–muscular coherence re-
garding their relation to clinical symptoms suggests that they are not in-
dependent, despite their different sensitivity tomovement and levodopa.
Indeed, we found a positive relationship between M1–STN and M1–
muscular beta coherence (Fig. S5). The latter was significant in HOLD
OFF (r=0.80, p=0.04) and observable as a trend in MOVE OFF (r=
0.65, p=0.08) and MOVE ON (r=0.72, p=0.06). Interestingly, it was
much weaker in HOLD ON (r=0.52, p=0.21), i.e. the experimental
condition in which the effect of levodopa administration on M1–STN
beta coherence was strongest.
Finally, we found indications that the negative relationship between
beta coherence and UPDRS scores may hold in the resting state as well
(Fig. S6). However, anti-correlationswereweaker in RESTOFF thandur-
ing task performance and did not reach significance (rM1_STN=−0.60,
p=0.12; rM1_MUSCLE=−0.68, p=0.07).
Effect of spatial sampling
We investigated coherence between a reference channel and cor-
tical ROIs which consisted of a single cortical location each. Thus, spa-
tial sampling of the cortical signal was very sparse. Spatial under-
sampling may act as confound, especially if the quality of the spatial
filter varies systematically across conditions. In order to assess the ro-
bustness of the results against changes in spatial sampling, we repeat-
ed the analysis with a different definition of the M1 ROI. Instead of a
single location, we considered a 2-dimensional grid in the axial plane
that was centered on the original M1 ROI (5×5 points, 5 mm spac-
ing). Importantly, changing the ROI did not lead to any qualitative
changes, neither with regard to modulation of coherence (Table S2),
nor with regard to correlations with UPDRS (Fig. S7).
Discussion
In this study, we investigated modulations of oscillatory coupling
between cortex, STN and muscle by movement and administration
of levodopa in akinetic-rigid PD patients. We found that beta band co-
herence between STN and motor cortex was reduced by administra-
tion of levodopa. Alpha and beta band coherence between forearm
muscle and motor cortex was strongly reduced during repetitive
movement compared to static contraction but was not affected by levo-
dopa. STN-cortical and cortico-muscular beta coherence were negative-
ly correlated with akinesia/rigidity UPDRS scores in the OFF state.
Methodological considerations
Before discussing the results in detail, we would like to point out
some methodological limitations. First, for organizational reasons
subjects were recorded the day after surgery. It has been shown
that electrode impedance is variable in the first days following sur-
gery (Rosa et al., 2010). Therefore, it is conceivable that LFP signal
quality and therewith coherence changed at a timescale of several
days after surgery. Given the slow dynamics of impedance changes,
they are, however, unlikely to have caused coherence changes across
experimental conditions which occurred within minutes. Moreover,
impedance changes are not expected to introduce couplings between
cortex and STN, i.e. to generate false positive couplings.
Second, we estimated contact position by projecting normalized co-
ordinates derived from post-surgical CT onto the Schaltenbrand–Wahren
atlas. Like all other available methods, the procedure is not precise
enough to know exactly whether a given contact was recording from
Fig. 5.M1–STN andM1–muscular beta coherence are anti-correlated with UPDRS scores in the OFF state. Beta coherence between motor cortex and STN is plotted against hemibody
akinesia/rigidity UPDRS OFF score in the upper row. The lower row shows the same plots for cortico-muscular coherence. Black lines indicate the best linear fit. Pearson's correlation
coefficients and the corresponding p-values are given in the figure headings.
209J. Hirschmann et al. / NeuroImage 68 (2013) 203–213
the STN or nearby areas. Atlas-based localization bears the advantage of
avoiding the potential risks and artifacts associated with post-surgical
MR imaging and the difficulty of identifying the STN on individual MR
images (O'Gorman et al., 2004; Videen et al., 2008). Strong support for
contact placement in the STN comes from the observed clinical effects
of stimulation and frommicroelectrode recordings during surgery rather
than from post-surgical imaging.
Third, the influence of changes in power and therewith signal-to-
noise ratio (SNR) is a fundamental problem in the analysis of coher-
ence differences (Palva and Palva, 2012; Schoffelen and Gross, 2009).
The statistical analysis of group data reduces the influence of this con-
found, as SNR variations need to be consistent across subjects in order
to have an effect. Furthermore, comparison of coherence between dif-
ferent signals can help to assess the impact of power changes. In this
study, STN-cortical and cortico-muscular coherence were modulated
differently, although the cortical signal was identical in both cases.
Thus, cortical power changes cannot explain the observed changes in
coherence. Power changes in the reference channel cannot explain
them either, as a power decrease (STN beta power in response to levo-
dopa) and a power increase (EMG beta power in response to move-
ment) were both associated with a coherence decrease. Nevertheless,
influences of power changes cannot be ruled out completely.
Fourth, we made use of a block design to investigate movement ef-
fects. Thus, signals were treated as stationary although EMG, LFP and
MEG signals are known to undergomovement-related changes. For ex-
ample, M1 beta power is reduced prior to movement and re-emerges
followingmovement execution (e.g. Pfurtscheller et al., 2003). In conse-
quence, the current study cannot answer howmucheach aspect of tem-
poral modulation contributed to the net effects reported here.
Finally, the possibility remains that LFP recordings contained contri-
butions from areas other than the STN. Amajor contribution is unlikely,
however, since bipolar referencing minimized the impact of volume
conduction. The fact that we did not find coherent sources in or near
the STN in a previous study (Hirschmann et al., 2011) evidences that
bipolar LFP channels record local STN activity which is not detected by
MEG.
Coherence with STG
Two studies independently reported resting state alpha coherence
between STN and temporal areas, such as STG (Hirschmann et al.,
2011; Litvak et al., 2011). Here we show that STN but not muscle activ-
ity is coherent with STG activity in the alpha band during motor tasks.
The functional significance of this coupling remains unclear. In this
study, coherence was neithermodulated bymovement nor by adminis-
tration of levodopa, suggesting that it does not play a role in the basal
ganglia-cortex sensorimotor loop. It is conceivable that alpha band co-
herence between STN and temporal cortex is related to auditory pro-
cessing (Airaksinen et al., 2011). Alternatively, one might speculate
that alpha band coherence reflects attentional processes, in which
alpha oscillations are thought to play an important role (Palva and
Palva, 2007). In fact, there are some indications for an involvement
of STG and STN in attention: in a functional MRI study designed to
reveal attention networks, presentation of an alerting, visual cue ac-
tivated STG and also the thalamus, which could mediate functional
connectivity with STN (Fan et al., 2005). In a study investigating
the neuropsychological effects of DBS it was found that, amongst
other effects, STN stimulation led to a decline in selective attention
(Smeding et al., 2006). Experimental manipulation of attention is re-
quired to further address the relevance of STN–STG alpha coherence
in attentional processes.
Effects of motor task
We measured coherence during repetitive movement and static
contraction. In line with previous literature (Baker et al., 1997; Kilner
et al., 1999, 2000), movement decreased M1–muscular beta coherence
compared to static contraction. M1–STN beta coherence, which has
Fig. 6. The negative correlation between coherence and UPDRS scores is frequency-specific. Scatter plots show M1-STN (upper row) and M1–muscular coherence (lower row) plot-
ted against hemibody akinesia/rigidity UPDRS OFF score. Coherence was averaged across HOLD OFF and MOVE OFF. Black lines indicate the best linear fit. Pearson's correlation co-
efficients and the according p-values are given in the figure headings.
210 J. Hirschmann et al. / NeuroImage 68 (2013) 203–213
been reported to be reduced during movement as well (Marsden et al.,
2001b), did not showamain effect ofmotor task in this study. However,
the levodopa-induced decrease in M1-STN beta coherence tended to be
weaker in the MOVE than in the HOLD condition, indicating a floor
effect.
In general, this study provides further evidence for a modulation
of STN-cortical and cortico-muscular coupling by movement and sup-
ports the general notion that oscillatory activity in the beta band is
suppressed during a change in motor state (Engel and Fries, 2010).
In addition, it shows that STN-cortical and cortico-muscular coher-
ence differ in their responsiveness to movement.
Recently, Litvak et al. (2012) investigated movement-induced
changes in STN-cortical coupling in detail and found short-lived in-
creases in the gamma band at the time of movement onset. In the
present study, effects in the gamma band were not observed on
the group level, most likely because their detection requires an event-
related experimental design. However, clear coherence peaks in the
gamma band were observed in a subset of subjects, consistent with
the findings by Litvak et al. and other previous studies (Cassidy et al.,
2002; Williams et al., 2002).
Effects of medication
We found that levodopa administration decreased beta coherence
between STN and motor cortex. This finding is in good keeping with a
study in the rat model of PD reporting that injection of the dopamine
receptor agonist apomorphine reduced STN-cortical beta coherence
in animals with a 6-hydroxydopamine (6-OHDA) midbrain lesion
(Sharott et al., 2005). Furthermore, it is in line with a study by
Williams et al. (2002), which investigated STN-cortical coherence
in PD patients before and after levodopa administration and presented
data evidencing a suppression of beta coherence by levodopa. The result
is not in agreement, however, with two more recent studies in PD pa-
tients which did not report such an effect (Lalo et al., 2008; Litvak et
al., 2011). These studies differed from the current study as they used a
different coupling measure (Lalo et al., 2008), investigated rest rather
than movement epochs (Litvak et al., 2011) and did not exclude
tremor-dominant PD patients. In this study, we excluded tremor-
dominant patients to assure that the observed changes in coherence
are not confounded by changes in tremor-associated coherence (Pollok
et al., 2008; Timmermann et al., 2003). Selection of tremor-free data
may also explain why we could not confirm the previously reported de-
crease in cortico-muscular alpha coherence and concurrent increase in
cortico-muscular beta coherence in response to levodopa administration
(Salenius et al., 2002), as these are characteristics of tremor alleviation
(Park et al., 2009; Wang et al., 2007).
Another plausible explanation why some previous studies did not
find an effect of levodopa on STN-cortical beta coherence could be that
the effect is mild and therefore hard to detect. In this study, we applied
amethodologywith strong emphasis on estimation reliability and statis-
tical power. We combined beamforming, multitaper spectral estimation
and ROI analysis. Beamforming offers protection against artifacts to
some extent (Litvak et al., 2010), making source-level estimates of co-
herence oftenmore reliable than sensor-level estimates. Estimation reli-
ability was further increased by using the multitapering method, i.e. by
averaging many estimates. Finally, the statistics computed from these
estimates did not have to be corrected for multiple comparisons since
we concentrated on only two ROIs. Thus, we were able to detect effects
which are easily overlooked with less sensitive approaches.
Is strong STN-cortical beta coherence pathological?
Beta oscillations in the STN have been labeled “antikinetic” (Brown,
2003) in the sense that strong beta activity is either correlated with or
causing the slowing of movement in PD. This view is supported by an
impressive body of evidence (Brown, 2007). Given that the cortex was
reported to drive STN activity in the beta band (Lalo et al., 2008; Litvak
et al., 2011;Williams et al., 2002), onemight expect STN-cortical beta co-
herence to reflect this pathological drive and consider it “antikinetic” as
well.
The present study casts doubt on the appropriateness of this label.
The negative correlation with UPDRS scores rather suggests that
strong M1-STN beta coherence may be beneficial for PD patients in
the OFF state. Interestingly, the negative relationship was observed
for both M1-STN and M1–muscular coherence in both motor tasks
and at rest, suggesting that it holds for baseline levels of beta band
coupling within a wider sensorimotor network.
The observed anti-correlation can be interpreted in two ways. One
could assume that PD patientswith a lowUPDRS score bearmore resem-
blance to healthy subjects with regard to STN-cortical coupling than pa-
tients with a higher score, i.e. that the observed anti-correlation can be
extrapolated. In this case, one would expect healthy humans, unlike
rats (Sharott et al., 2005), to show stronger M1–STN beta coherence
than subjects suffering from PD. In consequence, M1–STN beta coher-
encewould need to be considered a physiological parameterwhich is re-
duced in PD, aswas proposed forM1–muscular beta coherence (Salenius
et al., 2002). Hence, both couplings may reflect a pathologically reduced,
system-wide beta coupling baseline which is necessary for motor
control. This interpretation is in line with the observation that both
measures are correlated. However, it implies that levodopa reduces a
physiological coupling. A possible explanation for this seemingly para-
doxical effect could be that levodopa acts not by restoring the absolute
but the relative level of M1-STN beta coherence, thereby reestablishing
the balance among couplings in themotor system. Another possibility is
that the dopamine-induced reduction of M1-STN beta coherence is a
by-product of the dopamine-induced reduction of STN beta power.
This effect might create the misleading impression that the roles of
beta coherence and beta power in motor control are similar, although
theymay actually be fundamentally different. The latter idea is supported
by the observation that in some subjects M1-STN coherence and STN
power peak at different frequencies within the beta band.
As an alternative to viewing strong beta coherence as a part of
normal motor function, one may hypothesize that patients with a
low UPDRS score are compensating better for dopamine depletion
than patients with a higher score. M1-STN beta coherence may mark
the ability to compensate and be actively up-regulated in PD. Compensa-
tion may become oblivious when dopamine is administered, explaining
the reduction in coherence by administration of levodopa. This interpre-
tation would also explain why a correlation between coherence and
motor score was observed only in the OFF state. Supporting the idea of
cortex-driven compensation, motor cortex was shown to increase its os-
cillatory activity in the beta band (Degos et al., 2009) and its influence on
basal ganglia output structures in response to dopamine depletion in the
basal ganglia (Belluscio et al., 2007; Dejean et al., 2012; Magill et al.,
2001). The temporal evolution of changes in cortical beta band oscilla-
tions and cortico-basal ganglia coupling did not match the development
of motor impairments, suggesting that they relate to neuronal reorgani-
zation rather than causing or reflecting PD symptoms (Degos et al., 2009;
Dejean et al., 2012). It is possible that these changes serve compensation
and that STN-cortical beta coherence is a marker of cortical control
exerted on the STN.
Further studies are needed to test the hypotheses outlined above. A
better understanding of the underlyingmechanismsmay require inves-
tigation of not only beta oscillations but also their cross-frequency cou-
pling with activity in other frequency bands, such as high frequency
oscillations, which attracted interest recently (López-Azcárate etal., 2010; Özkurt et al., 2011).
Conclusion
We investigated oscillatory coupling in PD patients and found thatSTN-cortical and cortico-muscular coherence are correlated but can
211J. Hirschmann et al. / NeuroImage 68 (2013) 203–213
be modulated independently by levodopa, suggesting that they reflect
activity in two partly independent sub-loops within the motor system.
Being anti-correlated to akinesia and rigidity, they might reflect physi-
ological or compensatory rather than pathological mechanisms.
Role of the funding source
The funding source did neither influence the study design, collec-
tion, analysis or interpretation of data nor the decision to submit the
paper for publication.
Acknowledgments
The authors would like to express their sincere gratefulness to the
patients who participated in this study. Furthermore, we are very
thankful to Nienke Hoogenboom for helpful discussions of data analysis
and to the people of Medtronic (Dr. Ali Sarem-Aslani, Mr. Paul van
Venrooij, and Mr. Andreas Rolf) for technical support. In addition, we
thank Mrs. E. Rädisch for assistance with MRI scans. A.S. acknowledgessupport by the ERANET-Neuron Grant “PhysiolDBS” (Neuron-48-013)and by the DFG (FOR1328, SCHN 592/3-1). M.B. was supported by aMarie Curie Fellowship of the EU (FP7-PEOPLE-2009-IEF-253965).
Appendix A. Supplementary data
Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.neuroimage.2012.11.036.
References
Airaksinen, K., Mäkelä, J.P., Taulu, S., Ahonen, A., Nurminen, J., Schnitzler, A., Pekkonen, E.,2011. Effects of DBS on auditory and somatosensory processing in Parkinson's dis-ease. Hum. Brain Mapp. 32, 1091–1099.
Airaksinen, K., Butorina, A., Pekkonen, E., Nurminen, J., Taulu, S., Ahonen, A., Schnitzler, A.,Mäkelä, J.P., 2012. Somatomotor mu rhythm amplitude correlates with rigidity dur-ing deep brain stimulation in Parkinsonian patients. Clin. Neurophysiol. 123 (10)(October), 2010–2017.
Baker, S.N., Olivier, E., Lemon, R.N., 1997. Coherent oscillations in monkey motor cortexand hand muscle EMG show task‐dependent modulation. J. Physiol. 501, 225–241.
Belluscio, M.A., Riquelme, L.A., Murer, M.G., 2007. Striatal dysfunction increases basalganglia output during motor cortex activation in parkinsonian rats. Eur. J. Neurosci.25, 2791–2804.
Brown, P., 2003. Oscillatory nature of human basal ganglia activity: relationship to thepathophysiology of Parkinson's disease. Mov. Disord. 18, 357–363.
Brown, P., 2007. Abnormal oscillatory synchronisation in the motor system leads to im-paired movement. Curr. Opin. Neurobiol. 17, 656–664.
Brown, P., Oliviero, A., Mazzone, P., Insola, A., Tonali, P., Di Lazzaro, V., 2001. Dopaminedependency of oscillations between subthalamic nucleus and pallidum in Parkinson'sdisease. J. Neurosci. 21, 1033–1038.
Cassidy, M., Mazzone, P., Oliviero, A., Insola, A., Tonali, P., Di Lazzaro, V., Brown, P., 2002.Movement-related changes in synchronization in the human basal ganglia. Brain125, 1235–1246.
Degos, B., Deniau, J.-M., Chavez, M., Maurice, N., 2009. Chronic but not acute dopami-nergic transmission interruption promotes a progressive increase in cortical betafrequency synchronization: relationships to vigilance state and akinesia. Cereb.Cortex 19, 1616–1630.
Dejean, C., Nadjar, A., Le Moine, C., Bioulac, B., Gross, C.E., Boraud, T., 2012. Evolution ofthe dynamic properties of the cortex–basal ganglia network after dopaminergicdepletion in rats. Neurobiol. Dis. 46, 402–413.
Engel, A.K., Fries, P., 2010. Beta-band oscillations-signalling the status quo? Curr. Opin.Neurobiol. 20, 156–165.
Eusebio, A., Thevathasan, W., Doyle Gaynor, L., Pogosyan, A., Bye, E., Foltynie, T.,Zrinzo, L., Ashkan, K., Aziz, T., Brown, P., 2011. Deep brain stimulation can suppresspathological synchronisation in parkinsonian patients. J. Neurol. Neurosurg. Psychia-try 82, 569–573.
Fan, J., McCandliss, B.D., Fossella, J., Flombaum, J.I., Posner, M.I., 2005. The activation ofattentional networks. NeuroImage 26, 471–479.
Giannicola, G., Marceglia, S., Rossi, L., Mrakic-Sposta, S., Rampini, P., Tamma, F.,Cogiamanian, F., Barbieri, S., Priori, A., 2010. The effects of levodopa and ongoingdeep brain stimulation on subthalamic beta oscillations in Parkinson's disease.Exp. Neurol. 226, 120–127.
Goetz, C.G., Tilley, B.C., Shaftman, S.R., Stebbins, G.T., Fahn, S., Martinez-Martin, P.,Poewe, W., Sampaio, C., Stern, M.B., Dodel, R., et al., 2008. Movement DisorderSociety-sponsored revision of the Unified Parkinson's Disease Rating Scale (MDS-UPDRS): scale presentation and clinimetric testing results.Mov. Disord. 23, 2129–2170.
Gross, J., Kujala, J., Hamalainen, M., Timmermann, L., Schnitzler, A., Salmelin, R., 2001.Dynamic imaging of coherent sources: studying neural interactions in the humanbrain. Proc. Natl. Acad. Sci. U. S. A. 98, 694–699.
Halliday, D.M., Rosenberg, J.R., Amjad, A.M., Breeze, P., Conway, B.A., Farmer, S.F., 1995.A framework for the analysis of mixed time series/point process data—theory andapplication to the study of physiological tremor, single motor unit discharges andelectromyograms. Prog. Biophys. Mol. Biol. 64, 237–278.
Hammond, C., Bergman, H., Brown, P., 2007. Pathological synchronization in Parkinson'sdisease: networks, models and treatments. Trends Neurosci. 30, 357–364.
Hirschmann, J., Özkurt, T.E., Butz, M., Homburger, M., Elben, S., Hartmann, C.J., Vesper, J.,Wojtecki, L., Schnitzler, A., 2011. Distinct oscillatory STN-cortical loops revealed by si-multaneousMEG and local field potential recordings in patients with Parkinson's dis-ease. NeuroImage 55, 1159–1168.
Kilner, J.M., Baker, S.N., Salenius, S., Jousmäki, V., Hari, R., Lemon, R.N., 1999. Task-dependent modulation of 15-30 Hz coherence between rectified EMGs fromhuman hand and forearm muscles. J. Physiol. 516, 559–570.
Kilner, J.M., Baker, S.N., Salenius, S., Hari, R., Lemon, R.N., 2000. Human Cortical Mus-cle Coherence Is Directly Related to Specific Motor Parameters. J. Neurosci. 20,8838–8845.
Kühn, A.A., Williams, D., Kupsch, A., Limousin, P., Hariz, M., Schneider, G.-H., Yarrow, K.,Brown, P., 2004. Event-related beta desynchronization in human subthalamic nu-cleus correlates with motor performance. Brain 127, 735–746.
Kühn, A.A., Kupsch, A., Schneider, G.-H., Brown, P., 2006. Reduction in subthalamic 8-35Hz oscillatory activity correlates with clinical improvement in Parkinson's disease.Eur. J. Neurosci. 23, 1956–1960.
Kühn, A.A., Kempf, F., Brücke, C., Gaynor Doyle, L., Martinez-Torres, I., Pogosyan, A.,Trottenberg, T., Kupsch, A., Schneider, G.-H., Hariz, M.I., et al., 2008. High-frequencystimulation of the subthalamic nucleus suppresses oscillatory beta activity in pa-tients with Parkinson's disease in parallel with improvement in motor perfor-mance. J. Neurosci. 28, 6165–6173.
Lalo, E., Thobois, S., Sharott, A., Polo, G., Mertens, P., Pogosyan, A., Brown, P., 2008. Pat-terns of bidirectional communication between cortex and basal ganglia duringmovement in patients with Parkinson disease. J. Neurosci. 28, 3008–3016.
Lang, A.E., Lozano, A.M., 1998. Parkinson's Disease. N. Engl. J. Med. 339, 1130–1143.Levy, R., Ashby, P., Hutchison, W.D., Lang, A.E., Lozano, A.M., Dostrovsky, J.O., 2002. De-
pendence of subthalamic nucleus oscillations on movement and dopamine inParkinson's disease. Brain 125, 1196–1209.
Litvak, V., Eusebio, A., Jha, A., Oostenveld, R., Barnes, G.R., Penny, W.D., Zrinzo, L.,Hariz, M.I., Limousin, P., Friston, K.J., et al., 2010. Optimized beamforming for si-multaneous MEG and intracranial local field potential recordings in deep brainstimulation patients. NeuroImage 50, 1578–1588.
Litvak, V., Jha, A., Eusebio, A., Oostenveld, R., Foltynie, T., Limousin, P., Zrinzo, L., Hariz, M.I.,Friston, K., Brown, P., 2011. Resting oscillatory cortico-subthalamic connectivity inpatients with Parkinson's disease. Brain 134, 359–374.
Litvak, V., Eusebio, A., Jha, A., Oostenveld, R., Barnes, G., Foltynie, T., Limousin, P., Zrinzo, L.,Hariz, M.I., Friston, K., et al., 2012. Movement-Related Changes in Local and Long-Range Synchronization in Parkinson's Disease Revealed by Simultaneous Magneto-encephalography and Intracranial Recordings. J. Neurosci. 32, 10541–10553.
López-Azcárate, J., Tainta, M., Rodríguez-Oroz, M.C., Valencia, M., González, R., Guridi, J.,Iriarte, J., Obeso, J.A., Artieda, J., Alegre, M., 2010. Coupling between Beta and High-Frequency Activity in the Human Subthalamic Nucleus May Be a Pathophysiologi-cal Mechanism in Parkinson's Disease. J. Neurosci. 30, 6667–6677.
Magill, P., Bolam, J., Bevan,M., 2001. Dopamine regulates the impact of the cerebral cortexon the subthalamic nucleus–globus pallidus network. Neuroscience 106, 313–330.
Marsden, J., Limousin-Dowsey, P., Fraix, V., Pollak, P., Odin, P., Brown, P., 2001a.Intermuscular coherence in Parkinson's disease: effects of subthalamic nucleusstimulation. NeuroReport 12, 1113–1117.
Marsden, J.F., Limousin-Dowsey, P., Ashby, P., Pollak, P., Brown, P., 2001b. Subthalamicnucleus, sensorimotor cortex and muscle interrelationships in Parkinson's disease.Brain 124, 378–388.
Nolte, G., 2003. The magnetic lead field theorem in the quasi-static approximation andits use for magnetoencephalography forward calculation in realistic volume con-ductors. Phys. Med. Biol. 48, 3637–3652.
O'Gorman, R.L., Selway, R.P., Reid, C.J., Hotton, G.R., Hall, E., Jarosz, J.M., Polkey, C.E.,Hill, D.L.G., 2004. Registered computed tomography images as an alternative topostimplantation magnetic resonance imaging in the assessment of subthalamicelectrode placement. J. Comput. Assist. Tomogr. 28, 548–550.
Oostenveld, R., Fries, P., Maris, E., Schoffelen, J.-M., 2011. FieldTrip: Open Source Soft-ware for Advanced Analysis of MEG, EEG, and Invasive Electrophysiological Data.Comput. Intell. Neurosci. 2011, 1–9.
Özkurt, T.E., Butz, M., Homburger, M., Elben, S., Vesper, J., Wojtecki, L., Schnitzler, A.,2011. High frequency oscillations in the subthalamic nucleus: A neurophysiologi-cal marker of the motor state in Parkinson's disease. Exp. Neurol. 229, 324–331.
Palva, S., Palva, J.M., 2007. New vistas for alpha-frequency band oscillations. TrendsNeurosci. 30, 150–158.
Palva, S., Palva, J.M., 2012. Discovering oscillatory interaction networks with M/EEG:challenges and breakthroughs. Trends Cogn. Sci. (Regul. Ed.) 16, 219–230.
Park, H., Kim, J.S., Paek, S.H., Jeon, B.S., Lee, J.Y., Chung, C.K., 2009. Cortico-muscular co-herence increases with tremor improvement after deep brain stimulation inParkinson's disease. NeuroReport 20, 1444–1449.
Pfurtscheller, G., Graimann, B., Huggins, J.E., Levine, S.P., Schuh, L.A., 2003. Spatiotemporalpatterns of beta desynchronization and gamma synchronization in corticographicdata during self-paced movement. Clin. Neurophysiol. 114, 1226–1236.
Pollok, B., Makhloufi, H., Butz, M., Gross, J., Timmermann, L., Wojtecki, L., Schnitzler, A.,2008. Levodopa affects functional brain networks in parkinsonian resting tremor.Mov. Disord. 24, 91–98.
212 J. Hirschmann et al. / NeuroImage 68 (2013) 203–213
Pollok, B., Krause, V., Martsch, W., Wach, C., Schnitzler, A., Südmeyer, M., 2012. Motor-
cortical oscillations in early stages of Parkinson's disease. J. Physiol. 590, 3203–3212.Priori, A., Foffani, G., Pesenti, A., Tamma, F., Bianchi, A.M., Pellegrini, M., Locatelli, M.,
Moxon, K.A., Villani, R.M., 2004. Rhythm-specific pharmacological modulation ofsubthalamic activity in Parkinson's disease. Exp. Neurol. 189, 369–379.
Rosa, M., Marceglia, S., Servello, D., Foffani, G., Rossi, L., Sassi, M., Mrakic-Sposta, S.,Zangaglia, R., Pacchetti, C., Porta, M., et al., 2010. Time dependent subthalamic
local field potential changes after DBS surgery in Parkinson's disease. Exp. Neurol.
222, 184–190.Salenius, S., Avikainen, S., Kaakkola, S., Hari, R., Brown, P., 2002. Defective cortical
drive to muscle in Parkinson's disease and its improvement with levodopa.Brain 125, 491–500.
Saxe, R., Brett, M., Kanwisher, N., 2006. Divide and conquer: A defense of functional
localizers. NeuroImage 30, 1088–1096.Schaltenbrand, G., Wahren, W., 1977. Atlas For Stereotaxy of The Human Brain With
Guide to The Atlas For Stereotaxy of The Human Brain, Pkg. Thieme.Schoffelen, J.-M., Gross, 2009. Source connectivity analysis with MEG and EEG. Hum.
Brain Mapp. 30, 1857–1865.
Sharott, A., Magill, P.J., Harnack, D., Kupsch, A., Meissner, W., Brown, P., 2005. Dopaminedepletion increases the power and coherence of beta-oscillations in the cerebral cor-
tex and subthalamic nucleus of the awake rat. Eur. J. Neurosci. 21, 1413–1422.
Smeding, H.M.M., Speelman, J.D., Koning-Haanstra, M., Schuurman, P.R., Nijssen, P.,
van Laar, T., Schmand, B., 2006. Neuropsychological effects of bilateral STN stim-ulation in Parkinson disease: A controlled study. Neurology 66, 1830–1836.
Taulu, S., Simola, J., 2006. Spatiotemporal signal space separation method for rejectingnearby interference in MEG measurements. Phys. Med. Biol. 51, 1759–1768.
Taulu, S., Simola, J., Kajola, M., 2005. Applications of the Signal Space Separation Meth-od. IEEE Trans. Signal Process. 53, 3359–3372.
Thomson, D.J., 1982. Spectrum estimation and harmonic analysis. Proc. IEEE 70,
1055–1096.Timmermann, L., Gross, J., Dirks, M., Volkmann, J., Freund, H.-J., Schnitzler, A., 2003. The
cerebral oscillatory network of parkinsonian resting tremor. Brain 126, 199–212.Videen, T.O., Campbell, M.C., Tabbal, S.D., Karimi, M., Hershey, T., Perlmutter, J.S., 2008.
Validation of a fiducial-based atlas localization method for deep brain stimulation
contacts in the area of the subthalamic nucleus. J. Neurosci. Methods 168, 275–281.Wang, S., Chen, Y., Ding, M., Feng, J., Stein, J.F., Aziz, T.Z., Liu, X., 2007. Revealing the dy-
namic causal interdependence between neural and muscular signals in Parkinso-nian tremor. J. Franklin Inst. 344, 180–195.
Williams, D., Tijssen, M., Van Bruggen, G., Bosch, A., Insola, A., Di Lazzaro, V., Mazzone, P.,
Oliviero, A., Quartarone, A., Speelman, H., et al., 2002. Dopamine-dependent changesin the functional connectivity between basal ganglia and cerebral cortex in humans.
Brain 125, 1558–1569.
213J. Hirschmann et al. / NeuroImage 68 (2013) 203–213
J. Hirschmann et al. / NeuroImage 68 (2013) 203–213 supplementary. - page 1
Supplementary
Fig. S1: Average position of DBS electrode contacts. Stereotactic contact coordinates were normalized with respect to the distance between
anterior commissure (AC) and posterior commissure (PC), averaged and projected onto the
Schaltenbrand–Wahren atlas. Left: Axial slice 1.5mm ventral to MCP. Right: Coronal
slice 3mm caudal to mid-commissural point (MCP). Grid spacing is 9 mm; yellow crosses
depict standard deviations. Four subjects used their left hand in the motor task and
contributed to the average location in the right hemisphere. Six subjects used their right
hand and contributed to the average location in the left hemisphere.
J. Hirschmann et al. / NeuroImage 68 (2013) 203–213 supplementary. - page 2
Fig S2: Effects of levodopa and motor task on M1, STN and muscle power. Condition-specific means of M1, STN and muscle power. Power values were
log10-transformed and averaged across subjects. Error bars indicate the standard error
of the mean (SEM). Asterisks mark significant differences (p < 0.05).
J. Hirschmann et al. / NeuroImage 68 (2013) 203–213 supplementary. - page 3
Fig. S3: Movement reduces STN-muscular beta coherence.
A) Condition-specific means of STN-muscular coherence. Coherence values were
z-transformed and averaged across subjects. Error bars indicate the standard error of the
mean (SEM). Asterisks mark significant differences (p < 0.05).
B) Average STN-muscular coherence spectra. Shaded areas indicate SEMs.
J. Hirschmann et al. / NeuroImage 68 (2013) 203–213 supplementary. - page 4
Fig. S4: Correlation between pre-surgical UPDRS and beta coherence in the OFF state.
Beta coherence between M1 and STN was negatively correlated with pre-surgical hemibody
akinesia/rigidity UPDRS scores in the OFF state. HOLD OFF and MOVE OFF beta
coherence were averaged. Pre-surgical UPDRS scores from five days before surgery were
available solely for eight of ten subjects.
Fig. S5: Correlation between M1-STN and M1-muscular beta coherence. M1-STN and M1-muscular beta coherence were positively correlated in HOLD OFF.
Trends for a positive correlation were observed in MOVE OFF and MOVE ON.
J. Hirschmann et al. / NeuroImage 68 (2013) 203–213 supplementary. - page 5
Fig. S6: Correlation between UPDRS and resting state beta coherence in the OFF state. At rest, UPDRS and beta coherence showed the same negative relationship as observed
during motor task performance. However, the anti-correlations were not significant.i
Effect of Spatial Sampling
In order to exclude that spatial under-sampling of the cortical signal confounded the results,
we reproduced the main findings of this study with a different definition of the M1 ROI.
Instead of a single location, we considered a 2-dimensional grid of points with axial
orientation that was centered on the original M1 ROI (5x5 regular grid, 5 mm spacing). For
the first test, we averaged coherence over each point in the sensorimotor grid. Notably, the
modulations of coherence (Tab. S1, upper row) and the anti-correlation between beta
coherence and UPDRS scores in the OFF state (Fig. S7, left column) could be reproduced.
For the second test, we chose a location within the sensorimotor grid for each individual
subject. In analogy with ROI and reference channel selection in the original analysis, this
was the location with maximal STN-cortical beta coherence in REST OFF. Again, we
detected exactly the same modulations of coherence as with the original procedure (Tab. S1,
lower row). Moreover, the correlation coefficients quantifying the relationship between beta
coherence and UPDRS scores were very similar to the ones reported in the main article (Fig.
S7, right column). We conclude that the spatial sampling applied in the original analysis was
sufficient to capture the dominant activity in the sensorimotor region.
J. Hirschmann et al. / NeuroImage 68 (2013) 203–213 supplementary. - page 6
beta coherence M1-STN mean alpha coherence M1-muscle mean beta coherence M1-muscle mean
F p F p F p
medication 15.70 <0.01 medication 1.97 0.19 medication 0.24 0.64
motor task 3.35 0.10 motor task 9.90 0.01 motor task 19.68 <0.01
medication * motor task 3.79 0.08 medication * motor task 0.03 0.87 medication * motor task 0.06 0.81
beta coherence M1-STN max. alpha coherence M1-muscle max. beta coherence M1-muscle max.
F p F p F p
medication 17.76 <0.01 medication 2.67 0.17 medication 0.01 0.92
motor task 2.98 0.12 motor task 11.99 <0.01 motor task 13.17 <0.01
medication * motor task 1.91 0.20 medication * motor task <0.01 0.97 medication * motor task 0.10 0.76
Tab. S1: Effects of medication and movement on coherence are robust against changes in spatial
sampling. Beta coherence was measured between a reference channel and several locations within a
sensorimotor region of interest (ROI). Upper row: Beta coherence was averaged over locations
within the ROI. Lower Row: Beta coherence was extracted from the subject-specific location
within the ROI that showed maximal coherence in REST OFF. Significant effects are marked by
italics. Only signal pair – frequency band combinations with at least one significant effect are listed.
Fig. S7: The anti-correlation between beta coherence and UPDRS is robust against
changes in spatial sampling. The figure shows the relationship between beta coherence during motor task performance
(HOLD OFF and MOVE OFF averaged) and UPDRS OFF scores. Beta coherence was
measured between a reference channel and several locations within a sensorimotor region of
interest (ROI). Left column: Beta coherence was averaged over locations within the ROI.
Right column: Beta coherence was extracted from the subject-specific location within the
ROI that showed maximal coherence in REST OFF. i
i In the published version x- and y-label are erroneously swapped
BRAINA JOURNAL OF NEUROLOGY
A direct relationship between oscillatorysubthalamic nucleus–cortex coupling andrest tremor in Parkinson’s diseaseJan Hirschmann,1,2 Christian J. Hartmann,1,2 Markus Butz,1,3 Nienke Hoogenboom,1,2
Tolga E. Ozkurt,4 Saskia Elben,1,2 Jan Vesper,5 Lars Wojtecki1,2 and Alfons Schnitzler1,2
1 Institute of Clinical Neuroscience and Medical Psychology, Medical Faculty, Heinrich-Heine-University Dusseldorf, Dusseldorf, Germany
2 Department of Neurology, University Hospital Dusseldorf, Dusseldorf, Germany
3 Sobell Department of Motor Neuroscience and Movement Disorders, Institute of Neurology, University College London, London, UK
4 Department of Health Informatics, Informatics Institute, Middle East Technical University, Ankara, Turkey
5 Department of Functional Neurosurgery and Stereotaxy, University Hospital Dusseldorf, Dusseldorf, Germany
Correspondence to: Alfons Schnitzler
Institute of Clinical Neuroscience and Medical Psychology
Heinrich-Heine-University Dusseldorf
Universitatsstr. 1
D-40225 Dusseldorf
Germany
E-mail: [email protected]
Electrophysiological studies suggest that rest tremor in Parkinson’s disease is associated with an alteration of oscillatory activity.
Although it is well known that tremor depends on cortico-muscular coupling, it is unclear whether synchronization within and
between brain areas is specifically related to the presence and severity of tremor. To tackle this longstanding issue, we took
advantage of naturally occurring spontaneous tremor fluctuations and investigated cerebral synchronization in the presence and
absence of rest tremor. We simultaneously recorded local field potentials from the subthalamic nucleus, the magnetoencepha-
logram and the electromyogram of forearm muscles in 11 patients with Parkinson’s disease (all male, age: 52–74 years).
Recordings took place the day after surgery for deep brain stimulation, after withdrawal of anti-parkinsonian medication. We
selected epochs containing spontaneous rest tremor and tremor-free epochs, respectively, and compared power and coherence
between subthalamic nucleus, cortex and muscle across conditions. Tremor-associated changes in cerebro-muscular coherence
were localized by Dynamic Imaging of Coherent Sources. Subsequently, cortico-cortical coupling was analysed by computation
of the imaginary part of coherency, a coupling measure insensitive to volume conduction. After tremor onset, local field
potential power increased at individual tremor frequency and cortical power decreased in the beta band (13–30Hz). Sensor
level subthalamic nucleus-cortex, cortico-muscular and subthalamic nucleus-muscle coherence increased during tremor specif-
ically at tremor frequency. The increase in subthalamic nucleus-cortex coherence correlated with the increase in electromyogram
power. On the source level, we observed tremor-associated increases in cortico-muscular coherence in primary motor cortex,
premotor cortex and posterior parietal cortex contralateral to the tremulous limb. Analysis of the imaginary part of coherency
revealed tremor-dependent coupling between these cortical areas at tremor frequency and double tremor frequency. Our findings
demonstrate a direct relationship between the synchronization of cerebral oscillations and tremor manifestation. Furthermore,
they suggest the feasibility of tremor detection based on local field potentials and might thus become relevant for the design of
closed-loop stimulation systems.
doi:10.1093/brain/awt271 Brain 2013: 136; 3659–3670 | 3659
Received May 2, 2013. Revised July 4, 2013. Accepted August 2, 2013. Advance Access publication October 22, 2013
ß The Author (2013). Published by Oxford University Press on behalf of the Guarantors of Brain. All rights reserved.
For Permissions, please email: [email protected]
at Un
iversitaets- u
nd
Lan
desb
iblio
thek
Du
esseldo
rf on
Feb
ruary
10
, 20
14
http
://brain
.ox
ford
jou
rnals.o
rg/
Do
wn
load
ed fro
m
Keywords: Parkinson’s disease; tremor; magnetoencephalography; coherence; deep brain stimulation
Abbreviations: DBS = deep brain stimulation; ImCoh = imaginary part of coherency; LFP = local field potential;MEG = magnetoencephalography
IntroductionParkinson’s disease is a debilitating neurological disorder resulting
from progressive cell death of dopaminergic neurons in the mid-
brain (Lang and Lozano, 1998). Recent research revealed abnor-
mally strong synchronization of rhythmic neuronal activity both in
animal models of Parkinson’s disease and patients, suggesting that
Parkinson’s disease is associated with pathologically altered neur-
onal oscillations (Schnitzler and Gross, 2005; Hammond et al.,
2007). Enhanced synchronization was shown to play a role in
akinesia and rigidity (Kuhn et al., 2006) and was suggested to
be involved in parkinsonian tremor.
Tremor occurs in �75% of patients and may range from mild to
severe manifestations (Hoehn and Yahr, 1967; Hughes et al.,
1993). Classical parkinsonian tremor occurs at rest, and is attenu-
ated at movement onset (Deuschl et al., 2000). Therefore, it is
referred to as rest tremor. The frequency of parkinsonian rest
tremor ranges between 3 and 7Hz.
It is generally agreed that central rather than peripheral mech-
anisms underlie parkinsonian tremor (Elble, 1996; McAuley and
Marsden, 2000; Schnitzler et al., 2006). Currently, two overlap-
ping central networks are considered candidate generators: the
cerebello-thalamo-cortical circuit and the basal ganglia-cortical
motor loop (Helmich et al., 2012). Tremor-related neural activity
occurs in both networks, and lesions and deep brain stimulation
(DBS) of structures in either network lead to tremor suppression
(Bergman et al., 1990; Benabid et al., 1991; Krack et al., 1997).
Patient recordings from the ventrolateral thalamus revealed
coherence between single unit and muscle activity at tremor
frequency, suggesting that the thalamus is involved in tremor
generation (Lenz et al., 1988; Zirh et al., 1998). Moreover, the
ventral intermediate nucleus of the thalamus is considered the
most effective DBS target for tremor suppression (Deuschl et al.,
2000). As this nucleus receives mainly cerebellar afferents, it was
proposed that cerebellar activity also contributes to tremor expres-
sion (Stein and Aziz, 1999). In fact, imaging studies demonstrated
that DBS of the ventral intermediate nucleus affects cerebellar
blood flow and revealed that cerebellar blood oxygenation and
metabolic activity are positively correlated with tremor amplitude
(Deiber et al., 1993; Helmich et al., 2011; Mure et al., 2011).
Besides the cerebello-thalamic circuit, tremor research has focused
on the basal ganglia. Microelectrode recordings in non-human pri-
mates (Raz et al., 2000; Heimer et al., 2006) and patients undergo-
ing surgery (Hutchison et al., 1997) revealed so-called tremor cells
in the internal globus pallidus. These cells fire bursts at tremor fre-
quency and bursting is, at least transiently, coherent with tremor
recordings from the muscle (Hurtado et al., 2005).
Similar observations were made in the subthalamic nucleus. In
vervet monkeys, oscillations at tremor frequency and double
tremor frequency emerged when the animals began to develop
tremor due to 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine
(MPTP) injection (Bergman et al., 1994). Furthermore, power
spectra of subthalamic nucleus local field potentials (LFPs) show
peaks at tremor frequency and subthalamic nucleus LFPs are co-
herent with the EMG at tremor frequency (Levy et al., 2002; Liu
et al., 2002; Wang et al., 2005; Reck et al., 2009).
Notably, tremor-related oscillatory activity is not only found in
subcortical nuclei and the cerebellum, but also in the cortex.
Timmermann et al. (2003) studied cerebro-muscular coherence
using magnetoencephalography (MEG) and observed significant
coupling at tremor frequency and double the tremor frequency
in a network including primary motor cortex, premotor cortex,
posterior parietal cortex, cerebellum and a diencephalic source
that was assumed to be the thalamus. The same network was
later shown to underlie voluntary tremor in healthy controls
(Pollok et al., 2004), and a similar network was found to be
involved in essential tremor (Schnitzler et al., 2009).
In summary, several lines of evidence suggest that tremor mani-
festation is associated with cerebral oscillations at tremor fre-
quency, indicating that they could serve as a trigger signal in
closed-loop DBS (Rosin et al., 2011). The nature of this associ-
ation, however, remains elusive. Patient studies on subthalamic
nucleus single unit activity reported that rhythmic spiking around
5Hz can be observed in the absence of tremor (Magarinos-Ascone
et al., 2000; Moran et al., 2008; Shimamoto et al. 2013). These
results demonstrate that the presence of spectral peaks at tremor
frequency is not sufficient to make inferences on the tremor state.
Furthermore, they show that oscillations on the single cell level are
not sufficient to elicit tremor, suggesting that tremor manifestation
might require coordinated network activity.
In this study, we hypothesized that tremor depends on synchron-
ization within the motor system. Specifically, we aimed at demon-
strating that tremor is associated with modulations of subthalamic
nucleus power, cortical power, subthalamic nucleus-cortex and cor-
tico-cortical coupling. To this end, we simultaneously recorded sub-
thalamic nucleus LFPs, MEG and the EMG of forearm muscles in
tremor-dominant patients with Parkinson’s disease. As demonstrated
by several recent studies (Hirschmann et al., 2011, 2013; Litvak
et al., 2011, 2012; Oswal et al., 2013), this combination of record-
ing techniques is a powerful tool for studying connectivity between
subthalamic nucleus, cortex and muscle. We identified epochs of
spontaneous rest tremor as well as tremor-free epochs using the
EMG recordings and compared oscillatory activity across conditions.
The study critically extends our current knowledge about parkinso-
nian rest tremor by demonstrating the pivotal role of synchronous
oscillations in subthalamic nucleus and cortex.
Materials and methods
PatientsEleven patients with Parkinson’s disease who were clinically selected
for DBS because of levodopa-induced fluctuations and dyskinesias
3660 | Brain 2013: 136; 3659–3670 J. Hirschmann et al.
at Un
iversitaets- u
nd
Lan
desb
iblio
thek
Du
esseldo
rf on
Feb
ruary
10
, 20
14
http
://brain
.ox
ford
jou
rnals.o
rg/
Do
wn
load
ed fro
m
participated in this study with written informed consent. All subjects
suffered from moderate to severe rest tremor that was alleviated by
DBS (Table 1). Seven subjects showed bilateral tremor during the
recordings so that it was possible to include both hemispheres in the
analysis. Four subjects showed unilateral tremor so that we were re-
stricted to one hemisphere. Thus, 18 subthalamic nuclei were analysed
in total. Subject 6 had been included in an earlier study on akinesia
(Hirschmann et al., 2013). He showed transiently emerging tremor in
addition to severe akinesia and rigidity. The study was approved by
the local ethics committee (Study no. 3209) and is in accordance with
the Declaration of Helsinki.
SurgeryImplantation of electrodes was carried out at the Department of
Functional Neurosurgery and Stereotaxy of the University Hospital
Dusseldorf. The surgical procedures are described elsewhere (Ozkurt
et al., 2011). Oral anti-parkinsonian medication was withdrawn the
evening before surgery and substituted by subcutaneous apomorphine
medication. Eight of 11 subjects were implanted with electrode model
3389 (Medtronic Inc.). Subjects 6, 8 and 9 were implanted with a DBS
system by St. Jude Medical Inc. Electrode placement was guided by
intraoperative microelectrode recordings, intraoperative stimulation
and clinical testing of DBS efficacy.
Electrode contact localizationTo reconstruct the final electrode placement, preoperative MRIs and
postoperative CT scans were aligned using rigid transformation as
provided by the functional magnetic resonance imaging of the Brain
Linear Image Registration Tool (Jenkinson et al., 2012). Subsequently,
the electrode position was derived from its characteristic artefacts in
CT scans (Hemm et al., 2009). Contacts were labelled in a
0.5 � 0.5 � 0.5mm mask image in individual MRI space. For group
comparison, individual MRI scans were transformed to Montreal
Neurological Institute (MNI) space using the symmetric normalization
strategy implemented in Advanced Normalisation Tools (Avants et al.,
2008). The same transformation was applied to the mask images to
obtain contact positions in MNI space.
RecordingsWe simultaneously recorded LFPs from the subthalamic nucleus, MEG
and the EMG of the extensor digitorum communis and flexor digi-
torum superficialis muscles of both upper limbs. All recordings were
performed using a 306 channel, whole-head MEG system (Elekta Oy).
The sampling rate was 2000Hz. DBS electrodes were connected to the
amplifier integrated into the MEG system by non-magnetic extension
leads. Online filters were applied to create a passband of 0.03–660Hz
for MEG signals, and a passband of 0.1–660Hz for LFP and EMG
signals. EMG electrodes were referenced to surface electrodes at the
muscle tendons. DBS electrodes were referenced to a surface electrode
at the left mastoid and rearranged to a bipolar montage offline. Re-
referencing was performed by signal subtraction and yielded three
bipolar LFP channels per electrode: 0–1 (ventral), 1–2 and 2–3 (dorsal).
Clinical ratings and paradigmRecordings took place the day after surgery. Two hours before record-
ing apomorphine administration was stopped. The clinical OFF state
was quantified by means of the motor score of the Movement
Disorder Society Unified Parkinson’s Disease Rating Scale immediately
before the recording started (Goetz et al., 2008). The rating was per-
formed by an experienced movement disorders specialist. Inside the
Table 1 Clinical details of patients
Subject Gender Age(years)
Diseaseduration(years)
UPDRSIIIrecordingday
Individual tremorfrequency (hz)
Side OFF/OFF upperlimb rest tremorsubscore
OFF/ON upperlimb rest tremorsubscore
1 M 65 8 40 4.0 Right 1 0Left 3 2
2 M 69 6 51 3.5 Right 3 3Left 3 2
3 M 68 11 36 3.0 Left 1 0
4 M 59 6 39 4.5 Right 3 1Left 3 0
5 M 68 2 39 4.0 Right 1 0Left 2 1
6 M 52 11 31 6.0 Right 2 0Left 0 0
7 M 67 6 34 6.5 Right m.d. m.d.Left m.d. m.d.
8 M 53 12 26 5.0 Left 2 2
9 M 65 4 43 4.5 Right 4 0Left 4 0
10 M 74 7 60 5.0 Right 4 0
11 M 69 12 30 7.0 Right 3 0
Mean 64.45 7.73 39.00 4.82 2.40 0.60
Standard deviation 6.93 3.38 9.75 1.25 1.17 1.07
The column labelled side indicates which body sides were analysed. The last two columns show the effect of deep brain stimulation on upper limb rest tremor as
documented in the control assessment of motor symptoms �3 months after implantation of the stimulation device. OFF/ON signifies that medication was off and
stimulation was on (m.d. = missing data). M = male; UPDRS = Unified Parkinson’s Disease Rating Scale.
Oscillatory coupling and rest tremor Brain 2013: 136; 3659–3670 | 3661
at Un
iversitaets- u
nd
Lan
desb
iblio
thek
Du
esseldo
rf on
Feb
ruary
10
, 20
14
http
://brain
.ox
ford
jou
rnals.o
rg/
Do
wn
load
ed fro
m
shielded room, subjects were instructed to sit as still as possible with
eyes open. In the period analysed in this study, there was neither a
task nor any kind of stimulus presentation. The duration of the rest
recording varied according to the subsequent paradigm. Paradigm 1
(Subjects 1–8) included 10min of rest in total and is described in detail
in Hirschmann et al. (2013). Paradigm 2 (Subjects 9–11) included
3min of rest.
Epoch selectionRaw data were inspected by eye. Epochs were labelled as tremor epochs
if continuous 3–7Hz periodic activity was clearly apparent in the EMG
time series of both the upper limb extensor and flexor. Simultaneous
tremor on the contralateral body side was not accounted for when
determining tremor epochs, i.e. we did not differentiate between bilat-
eral and unilateral tremor. Epochs were labelled as tremor-free only if
neither forearm showed any periodic activity. Epochs containing arte-
facts such as contraction of jaw muscles or coughing were discarded.
Data selection yielded 82 s of tremor-free data (range: 17–235 s) and
63 s of tremor data (range: 14–201 s) on average.
PreprocessingTemporal Signal Space Separation (Taulu and Simola, 2006) was
applied using MaxFilter (Elekta Oy) as a means to shield the MEG
signal from tremor-related muscle activity. A discrete Fourier transform
filter was applied to remove any remaining power line noise (50Hz)
and its first two harmonics (100 and 150Hz). This processing step and
all of the following were performed using Matlab R2012a (The
Mathworks) and the FieldTrip toolbox (Oostenveld et al., 2011).
EMGs were high-pass filtered at 10Hz and full-wave rectified. Data
were down-sampled to 256Hz.
Channel selectionA set of 24 gradiometers contralateral to the tremulous limb was se-
lected a priori as MEG sensors of interest. The sensors were chosen
such that they covered sensorimotor and premotor motor cortex
(Fig. 1A). The selection was adapted for each body side individually: in
the sensors of interest, power was averaged over the individual tremor
frequency band (tremor frequency �0.5Hz) and the first harmonic
band (double tremor frequency �0.5Hz) and summed over conditions
(tremor and tremor-free episodes). Subsequently, the sensor with
maximum power and its six nearest neighbours were selected.
Furthermore, one LFP and one EMG channel were selected for each
body side. For each of the three LFP channels contralateral to the
tremulous limb, we computed LFP-MEG coherence and averaged
across MEG channels of interest, resulting in one spectrum per LFP
channel and condition. Coherence spectra from both conditions were
summed, and we selected the LFP channel with highest coherence at
individual tremor frequency for further analysis. Selection of the EMG
channel of interest was performed analogously (either the forearm
extensor or the forearm flexor of the tremulous limb was chosen).
Fig. 1B shows the positions of selected LFP channels in MNI space
together with a probability map of the subthalamic nucleus (Forstmann
et al., 2012).
Sensor level analysis
Time-frequency representations
To investigate the dynamics of tremor-related LFP and MEG power at
tremor onset, we selected all available tremor epochs with a discernible
tremor onset that lasted 510 s and were separated from the previous
tremor epoch of the same limb by 510 s. Time-frequency represen-
tations were produced by Fourier transformation of Hanning-tapered
data in a sliding window that was moved in steps of 50ms. Window
length was set to 2 s for 1–4.5Hz and to seven cycles for 5–30Hz to
obtain a better time resolution.
For statistical analysis, time-frequency representations were aligned
to individual tremor frequency and compared to baseline (ÿ9 to 0 s
relative to tremor onset) using a non-parametric, cluster-based ran-
domization approach (Maris and Oostenveld, 2007). In case multiple
epochs were available for a single subthalamic nucleus, the corres-
ponding time-frequency representations were averaged prior to statis-
tical analysis. In short, a group statistical image was computed and
two thresholds were applied. In this case, the thresholds were chosen
to be the 0.05 and 0.95 percentiles of the distribution of the ‘activa-
tion versus baseline t-value’. Following threshold application, values of
neighbouring supra-threshold voxels were summed and the cluster
sums were stored. Then, subject-specific images were randomly
shuffled across conditions, an alternative statistical image was com-
puted and cluster sums were computed as before. By repeating this
step 1000 times, an empirical, non-parametric null distribution was
constructed to which the original cluster sums were compared.
Importantly, this approach effectively controls for multiple compari-
sons. Please note, however, that it does not account for possible stat-
istical dependencies between hemispheres.
Figure 1 Location of selected channels. (A) A priori MEG sensor selection for subjects with left upper limb tremor. (B) Location of the
selected LFP channels (red dots) in MNI space. The blue cloud represents a subthalamic nucleus probability map (Forstmann et al., 2012).
Blue voxels belong to the subthalamic nucleus with a probability of 44%. to belong to the subthalamic nucleus. Left: Coronal slice at
y = ÿ17.5mm seen from anterior. Right: Axial slice at z = ÿ5.5mm seen from inferior.
3662 | Brain 2013: 136; 3659–3670 J. Hirschmann et al.
at Un
iversitaets- u
nd
Lan
desb
iblio
thek
Du
esseldo
rf on
Feb
ruary
10
, 20
14
http
://brain
.ox
ford
jou
rnals.o
rg/
Do
wn
load
ed fro
m
Coherence spectra
For coherence analysis, data were divided into half-overlapping segments
of 2-s length. Subsequently, data segments were arranged into two sep-
arate sets containing tremor and tremor-free episodes, respectively.
Segments were convolved with a Hanning taper and coherence was
calculated for both conditions. As in the analysis of power, coherence
spectra were aligned to individual tremor frequency and compared across
conditions using a non-parametric, cluster-based randomization approach.
The dependent samples t-value served to define the cluster thresholds.
Correlation between EMG power and LFP-MEG coherence was
quantified by Pearson’s linear correlation coefficient. Power was con-
sidered in logarithmic units and coherence was Fisher z-transformed.
In order to account for peaks at tremor frequency and at its first
harmonic, we averaged coherence over the tremor frequency band
and the first harmonic band prior to computing correlation.
Source analysisSource analysis was performed using beamforming, a spatial filtering
approach. Importantly, tremor and tremor-free epochs were both pro-
jected through a common, real-valued spatial filter that was derived
from the joint data from both conditions. This step excludes the pos-
sibility that statistical differences between conditions occur due to dif-
ferences in spatial filters.
Subthalamic nucleus-cortex and cortico-muscularcoherence
Estimation of subthalamic nucleus-cortex and cortico-muscular coher-
ence on the source level was realized by Dynamic Imaging of
Coherent Sources (Gross et al., 2001). Regularization was set to 5%
of the mean of the trace of the channel cross-spectral density matrix.
Source orientation was defined as the orientation that maximized power.
The forward model was based on a realistic, single shell head model
derived from individual T1-weighted structural MRIs (Nolte, 2003). The
latter were obtained prior to surgery using a Magnetom Trio MRI scan-
ner (Siemens). We made use of regular beamformer grids with 5mm
spacing that were aligned to MNI space (Mattout et al., 2007). All
analysed beamformer grid points lay within 1.5 cm from the cortical
surface, i.e. we did not consider subcortical structures. Limiting the ana-
lysis to cortical areas served to increase statistical power.
Statistical analysis of source level coherence was performed in the
same way as for sensor level coherence. The non-parametric random-
ization approach is suited to analyse one-dimensional input, such as
coherence spectra, as well as multi-dimensional input such as volumet-
ric images (Maris and Oostenveld, 2007). A one-sided test was used as
we explicitly sought to localize the coherence increases observed in the
previous sensor level analysis.
Cortico-cortical coupling
For investigation of cortico-cortical coupling, the time domain activity
of selected sources was reconstructed using a Linear Constraint
Minimum Variance beamformer (Van Veen and Buckley, 1988).
Regularization was set to 20%. To improve the signal-to-noise ratio,
we made use of the FieldTrip implementation of the eigenspace beam-
former approach (Sekihara et al., 2002). In this approach, a projection
onto a subspace of the data covariance matrix is applied to remove
noise components. We chose the subspace spanned by the first N
singular vectors of the sensor covariance matrix with corresponding
singular values �1 to �N, such that �i /�14 0.2 for all i � [1, 2, . . . , N].
Following reconstruction of source time courses, we calculated coher-
ence and the imaginary part of coherency (ImCoh) for all source pairs.
ImCoh is a coupling measure related to coherence. Unlike coherence, it
is unaffected by volume conduction and therefore better suited to in-
vestigate cortico-cortical coupling (Nolte et al., 2004). To improve the
signal-to-noise ratio, the data were convolved with three Slepian tapers
before analysing cortico-cortical coupling (Thomson, 1982).
Cortico-cortical coupling was statistically analysed using a repeated-
measures ANOVA. The non-parametric randomization approach was
not applied in this case since there is no established procedure to
assess the influence of multiple factors. As in correlation analysis, co-
herence and ImCoh were averaged over the tremor frequency band
and the first harmonic band. Coherence was Fisher z-transformed and
ImCoh was rectified. The latter step ensured that subject-specific
ImCoh values did not cancel in the group average. ANOVAs included
the factors ‘tremor’ (no tremor, tremor), ‘pair’ (source X ÿ source Y,
source X ÿ source Z, . . .) and ‘shuffling’ (original, shuffled). In the
shuffled condition, one signal in each pair was shifted forward in
time by k segments (circular shift). k was a random integer between
2 and M ÿ 1, M being the total number of segments. We applied
Greenhouse-Geisser correction for non-sphericity where appropriate.
ResultsFor each subject, we determined the individual tremor frequency.
The latter was defined as the frequency of the first clear peak in
the EMG power spectrum during tremor. Additional peaks at
tremor frequency harmonics were observed in 15 cases.
Individual tremor frequencies ranged between 3 and 7Hz
(Table 1). In every subject, individual tremor frequency was con-
sistent across EMGs from different muscles and limbs.
Sensor level power
We investigated changes in MEG and LFP power around tremor
onset. Twenty-nine tremor epochs from 17 subthalamic nuclei and
10 subjects were included in this analysis. One subject was
excluded because tremor onset could not be determined. For
seven subthalamic nuclei, more than one epoch was available
(average: 2.7, range: 2–6). In these cases, subject-specific time-
frequency representations were averaged prior to statistical ana-
lysis (see ‘Materials and methods’).
Figure 2A shows an example of tremor onset (Subject 2, right
subthalamic nucleus). In this case, tremor amplitude did not in-
crease linearly over time. Instead, tremor developed in a staged
fashion. Stage transitions were reflected by MEG power decreases
in the beta band, followed by beta power increases and increases
at tremor frequency. An enhancement of LFP power at tremor
frequency occurred only in the last stage, when tremor amplitude
was maximal.
Figure 2B depicts the time course of group level MEG and LFP
power statistically compared to baseline (ÿ9 to 0 s). Please note that
time-frequency representations were aligned to individual tremor fre-
quency (f). Following tremor onset, both MEG and LFP power
showed a similar pattern: narrow-band power increases at tremor
frequency and its first harmonic co-occurred with a decrease in the
higher frequencies, corresponding to the beta band (13–30Hz). For
LFP power, the increase at tremor frequency reached significance
4.8 s after tremor onset (P = 0.01). MEG power decreased signifi-
cantly between 10 and 20Hz relative to tremor frequency in the
Oscillatory coupling and rest tremor Brain 2013: 136; 3659–3670 | 3663
at Un
iversitaets- u
nd
Lan
desb
iblio
thek
Du
esseldo
rf on
Feb
ruary
10
, 20
14
http
://brain
.ox
ford
jou
rnals.o
rg/
Do
wn
load
ed fro
m
period from 8–9 s after tremor onset (P50.01). Investigation of the
original, non-aligned time-frequency representations revealed a cor-
responding power decrease between 15 and 25Hz (P = 0.01; data
not shown). As depicted in Supplementary Fig. 1, this effect was due
to a sustained beta power suppression that began around tremor
onset and intensified as tremor continued. Group average LFP beta
power decreased between ÿ4 and 0 s and increased transiently at
tremor onset (Fig. 2B).
Sensor level coherence
As depicted in Fig. 3A, alignment of coherence spectra to individ-
ual tremor frequency revealed tremor-related coherence increases
between all pairs of signals specifically at tremor frequency
(LFP-MEG: P = 0.02, EMG-MEG: P50.01, EMG-LFP: P = 0.04).
Notably, the tremor-induced change in EMG power was positively
correlated with the change in LFP-MEG coherence (r = 0.50,
P = 0.03; Fig 3B). We did not test for correlations between EMG
power and LFP-EMG or EMG-MEG coherence since in these cases
power changes are likely to cause coherence changes, leading to
trivial correlations.
Source level coherence
The sensor level results show that coupling at tremor frequency
between subthalamic nucleus, cortex and muscle increases
Figure 2 Cortex and subthalamic nucleus showed tremor-related power changes. (A) Exemplary data from a tremor phase in Subject 2.
The figure shows the rectified EMG signal (top), MEG power (middle) and LFP power (bottom) around tremor onset (dotted line). MEG
power was averaged over the sensors of interest. Time-frequency plots were baseline-corrected (baseline: ÿ4 to 0 s) and the relative
power change is colour-coded. Note that sudden increases of tremor amplitude are preceded by MEG beta power decreases and followed
by power increases at tremor frequency. (B) Group statistical image of MEG and LFP power showing the contrast between the period from
0 to 9 s and the baseline period (ÿ9 to 0 s). Time-frequency representations were aligned to individual tremor frequency (f) i.e. individual
time-frequency representations were shifted along the frequency axis until the individual tremor frequency reached the 0Hz position.
Significant effects are highlighted by increased colour intensity (P5 0.05; n = 17) and t-values are colour-coded. Top: MEG power
between 10 and 20Hz relative to individual tremor frequency (f + 10 ÿ f + 20) decreased gradually. The effect was significant between 8
and 9 s after tremor onset. Bottom: Starting �1 s after tremor onset, LFP power increased at tremor frequency (f). The effect was
significant from 4.8 to 9 s after tremor onset.
3664 | Brain 2013: 136; 3659–3670 J. Hirschmann et al.
at Un
iversitaets- u
nd
Lan
desb
iblio
thek
Du
esseldo
rf on
Feb
ruary
10
, 20
14
http
://brain
.ox
ford
jou
rnals.o
rg/
Do
wn
load
ed fro
m
when tremor occurs. To identify the brain areas involved in this
process, we computed source level coherence at tremor frequency
and contrasted tremor and tremor-free epochs. As the quality
of spatial filters depends on the amount of data, we
included only subthalamic nuclei for which at least 30 s of rest
tremor and 30 s of tremor-free episodes were available. The
inclusion criterion was met by eight subthalamic nuclei from six
subjects.
Source level analysis revealed significant changes in cortico-mus-
cular but not in subthalamic nucleus-cortex coherence. One cluster
was found (P = 0.02; Fig. 4), which covered several motor and
sensory areas contralateral to tremor. By visual inspection, we
identified three local maxima that appeared to be distinct sources.
They were located in the primary motor cortex (MNI coordinates:
�60, ÿ15, 50), premotor cortex (MNI coordinates: �30, 10, 70)
and posterior parietal cortex (MNI coordinates: �20, ÿ75, 50).
Supplementary Fig. 2 shows the spatial configuration of selected
sources in detail.
Cortico-cortical coupling
To test whether the identified cortical sources are themselves
coupled, we estimated their time domain activity and computed
ImCoh for all source pairs. In line with previous reports
(Timmermann et al., 2003; Pollok et al., 2004), we found
cortico-cortical coupling to occur not only at tremor frequency,
but also and more frequently at double the tremor frequency.
Two representative examples of cortico-cortical coupling are
shown in Fig. 5.
A repeated-measures ANOVA with factors ‘shuffling’ and ‘pair’
revealed that shifting one signal in time destroyed ImCoh at
tremor frequency and its first harmonic. A main effect of shuffling
was found when considering tremor epochs [F(1,7) = 5.95,
P50.05] and a trend was observed for tremor-free epochs
[F(1,7) = 4.65, P = 0.07]. We found neither a main effect of pair
[tremor: F(2,14) = 0.11, P = 0.83; no tremor: F(2,14) = 0. 04,
P = 0.94] nor an interaction between shuffling and pair [tremor:
F(2,14) = 0.58, P = 0.57; no tremor: F(2,14) = 0. 19, P = 0.78].
Rather than affecting coupling between specific pairs of cortical
areas, shuffling reduced ImCoh between all pairs to a similar
degree (Fig. 6A).
Tremor-related changes incortico-cortical coupling
The fact that ‘shuffling’ had a slightly stronger effect in the tremor
condition might hint at an influence of tremor on ImCoh. This
possibility was not investigated further because a conditional
change in ImCoh is difficult to interpret. It may be explained
either by a change in phase consistency or by alteration of the
Figure 3 Subthalamic nucleus, cortical motor areas and muscle
synchronized during tremor. (A) Plots show mean LFP-MEG,
EMG-MEG and LFP-EMG coherence in the presence (red) and
absence of tremor (blue). Spectra were aligned to individual
tremor frequency (f) before averaging. Coherence with MEG
was averaged over the sensors of interest. Black, horizontal bars
indicate significant differences (P50.05; n = 18). Shaded areas
indicate standard error of the mean. (B) Changes in LFP-MEG
coherence are plotted against changes in EMG power. The line
indicates the best linear fit. Values were averaged over the
tremor frequency and its first harmonic.
Figure 4 The muscle coupled to a distributed sensorimotor
network during tremor. Surface plot illustrates the difference in
cortico-muscular coherence between tremor and tremor-free
epochs. t-values are colour-coded and only significant changes
at individual tremor frequency are displayed (P50.05; n = 8).
Images depicting right hand tremor (five of eight subthalamic
nuclei) were mirrored across the mid-sagittal plane so that the
right hemisphere can be considered the hemisphere contralateral
to the tremulous limb.
Oscillatory coupling and rest tremor Brain 2013: 136; 3659–3670 | 3665
at Un
iversitaets- u
nd
Lan
desb
iblio
thek
Du
esseldo
rf on
Feb
ruary
10
, 20
14
http
://brain
.ox
ford
jou
rnals.o
rg/
Do
wn
load
ed fro
m
preferred phase difference (Gross et al., 2013). As coherence is
unaffected by a change in the preferred phase difference, we used
coherence to quantify the difference between tremor and tremor-
free epochs.
An ANOVA with factors ‘tremor’ and ‘pair’ revealed a main
effect of tremor [F(1,7) = 7.48, P = 0.03] and a main effect of
pair [F(2,14) = 5.65, P = 0.04] but no interaction between tremor
and pair [F(2,14) = 1.0, P = 0.36]. As depicted in Fig. 6B, coher-
ence between all pairs increased in the tremor condition.
The effect of pair was most likely due to volume conduction, as
neighbouring areas exhibited higher coherence than distant areas,
regardless of the tremor state.
DiscussionWe investigated the parkinsonian rest tremor network by means
of simultaneous LFP, MEG and EMG recordings and found that
cerebral synchronization at tremor frequency increases as tremor
becomes manifest. Increases were observed in a network including
subthalamic nucleus, primary motor, premotor and posterior par-
ietal cortex contralateral to the tremulous limb. In addition, we
demonstrated that the tremor-associated increase in subthalamic
nucleus-cortex coherence was positively correlated with the
tremor-associated increase in muscle activity.
Methodological considerations
Simultaneous LFP, MEG and EMG recordings provide unique in-
sights into the relationship between subcortical, cortical and
muscle activity in humans and enabled recent advances in the
characterization of functional connectivity in Parkinson’s disease
(Hirschmann et al., 2011, 2013; Litvak et al., 2011, 2012;
Oswal et al., 2013). Before discussing the results of the current
study in detail, we will consider some methodological aspects
associated with coherence measurements obtained with this
approach.
One of the major concerns in coherence analysis is the possibil-
ity that changes in coherence are trivial consequences of changes
in power (Schoffelen and Gross, 2009). Although coherence is
normalized by power, it is affected by power changes as they
result in changes in the signal-to-noise ratio (Palva and Palva,
2012). In tremor analysis, the most drastic power changes occur
in the EMG. Therefore, one might expect EMG power changes to
cause changes in cortico-muscular coherence. Importantly, group
statistical analysis on the source level excluded this potential con-
found by demonstrating consistent spatial patterns. Changes in
cortico-muscular coherence were consistently observed in a limited
set of cortical areas contralateral to the tremulous limb. There is no
plausible mechanism by which EMG power changes might affect
specifically these areas while sparing all others.
Apart from affecting the signal-to-noise ratio, tremulous move-
ment creates rhythmically changing magnetic fields that may in
principle be measured by MEG directly, resulting in artefacts and
spurious cortico-muscular coherence. A systematic effect of the
latter seems unlikely for the same reasons that speak against con-
founds because of EMG power changes. Artefacts were rarely
observed in this study due to the usage of non-magnetic exter-
nalization leads and application of temporal signal space
separation.
Finally, imprecise determination of tremor onset could have
influenced the results on power time courses. As the emergence
of tremor was often gradual, it was not always possible to deter-
mine the exact moment of tremor onset. This inevitable impreci-
sion diminished the detection probability of transient effects, since
detection requires a high degree of temporal overlap across
Figure 6 Cortico-cortical coupling increased during tremor.
(A) Bars represent mean, absolute ImCoh between pairs of
cortical sources during tremor (n = 8). (B) Bars represent mean,
z-transformed coherence between pairs of cortical sources.
Values were averaged over the tremor frequency and its first
harmonic. Error bars indicate the standard error of the mean.
The y-axis scale is the same for all sub-plots within one row.
M1 = primary motor cortex; PPC = parietal cortex;
PMC = premotor cortex.
Figure 5 Cortical areas in the tremor network are coupled
to one another at tremor frequency and/or double tremor
frequency. Plots show examples of ImCoh between pairs of
cortical sources from Subject 6 (top row) and Subject 2 (bottom
row). Blue = no tremor; red = tremor; black = shuffled. Vertical
lines indicate the tremor frequency and its first harmonic.
M1 = primary motor cortex; PPC = parietal cortex;
PMC = premotor cortex.
3666 | Brain 2013: 136; 3659–3670 J. Hirschmann et al.
at Un
iversitaets- u
nd
Lan
desb
iblio
thek
Du
esseldo
rf on
Feb
ruary
10
, 20
14
http
://brain
.ox
ford
jou
rnals.o
rg/
Do
wn
load
ed fro
m
epochs. Thus, the analysis was biased towards the detection of
sustained effects. Moreover, detection probability increased with
time after tremor onset, as any jitter in onset times affected the
temporal overlap of sustained effects only in the first seconds after
tremor onset.
Subthalamic nucleus and cortical power
In line with previous studies (Levy et al., 2002; Liu et al., 2002;
Wang et al., 2006; Reck et al., 2009), we observed clear peaks in
the subthalamic nucleus LFP power spectra at tremor frequency
and double tremor frequency. Moreover, we found a tremor-
induced increase of subthalamic nucleus power at individual
tremor frequency, as reported in a previous single case study
(Wang et al., 2005). Notably, the increase occurred several
seconds after tremor onset, at a time when tremor amplitude
had reached its maximum. Hence, this power increase cannot be
the cause of tremor, but could reflect a gradual entrainment of
more and more subthalamic nucleus neurons, e.g. by sustained
somatosensory feedback (Wang et al., 2007; Florin et al.,
2010). Alternatively, it is conceivable that the detected subthala-
mic nucleus activity is related to the scale of tremor. Given that
the basal ganglia are important for movement scaling (Oliverira
et al., 1998; Desmurget and Turner, 2010), strong and sustained
subthalamic nucleus oscillations might emerge only when tremor
amplitude exceeds a certain threshold, i.e. when a large-scale
movement is being executed.
In addition to changes at tremor frequency, we found that cor-
tical beta power was suppressed during continuous tremor. This
finding tallies with a study in healthy subjects that reported a
sustained decrease of cortical beta power during repetitive, volun-
tary movement (Erbil and Ungan, 2007). Thus, our results further
strengthen the claim that voluntary movement and tremor have a
common neurophysiological basis (Pollok et al., 2004; Schnitzler
et al., 2006). However, they also provide indications for differ-
ences with respect to the dynamics of power. In this study, we did
not observe a decrease of cortical beta power before tremor onset,
whereas this effect is known to occur before voluntary movement
(Pfurtscheller et al., 2003). Further studies are needed to elaborate
on these potential differences in cortical activity.
Subthalamic nucleus-cortex coherence
While coherence between subthalamic nucleus and EMG at tremor
frequency has been addressed by numerous studies (Wang et al.,
2006; Amtage et al., 2008; Reck et al., 2009, 2010), subthalamic
nucleus-cortex coupling has rarely been investigated in the context
of tremor. Importantly, we demonstrated that subthalamic nu-
cleus-cortex coherence increases in the presence of tremor and
correlates with tremor severity, showing that: (i) the subthalamic
nucleus is part of the central tremor network; and (ii) it generates
input to cortex or receives output from cortex that directly reflects
tremor amplitude. These findings are complemented by a recent
intraoperative study reporting that phase-locking of subthalamic
nucleus spikes to motor cortical 6Hz oscillations is more
common in the presence than in the absence of tremor
(Shimamoto et al., 2013).
The paucity of epochs did not allow for localization of the
tremor-associated increase in subthalamic nucleus-cortex coher-
ence observed on the sensor level. Many subjects showed either
continuous tremor intermitted by short breaks or short episodes of
tremor, resulting in limited amounts of data suited for balanced
contrasts and spatial filter construction. Although the amount of
epochs sufficed to localize the change in cortico-muscular coher-
ence, localization of the weaker change in subthalamic nucleus-
cortex coherence likely requires more or longer recordings.
Cortico-muscular and cortico-corticalcoherence
In keeping with previous studies (Volkmann et al., 1996; Hellwig
et al., 2000), we found strong cortico-muscular coherence at
tremor frequency and its first harmonic. Coherence increased
during epochs of spontaneously emerging rest tremor and the in-
crease could be localized to a set of cortical areas contralateral to
the tremulous limb. The fact that increases occurred in both motor
and sensory cortical areas suggests that both efferent and afferent
rhythmical signalling is enhanced during tremor.
The localization presented in this study closely resembles the
tremor network identified in previous MEG (Timmermann et al.,
2003; Pollok et al., 2009) and EEG studies (Muthuraman et al.,
2012) on parkinsonian tremor. Thus, there is mounting evidence
for the existence of a cortical network including primary motor,
premotor and posterior parietal cortex that is active during patho-
logical and voluntary tremor (Pollok et al., 2004). Interestingly, a
recent study revealed the therapeutic potential of modulating the
cortical tremor network (Brittain et al., 2013). The study demon-
strated that interfering with cortical oscillations by transcranial
alternating current stimulation over motor cortex leads to substan-
tial tremor alleviation.
In line with the aforementioned studies (Timmermann et al.,
2003; Pollok et al., 2004, 2009), we found that the cortical
areas coherent with muscle activity are also coupled to one an-
other. The current results additionally show that cortico-cortical
coupling at tremor frequency is dependent on tremor manifest-
ation and is not a trivial consequence of volume conduction.
Comparison with previous studies
Earlier mappings of tremor-related coherence led to the identifica-
tion of more areas than reported in this study (Timmermann et al.,
2003; Pollok et al., 2004, 2009; Muthuraman et al., 2012). In
addition to primary motor, premotor and posterior parietal
cortex, significant coherence was observed in the supplementary
motor area, secondary somatosensory cortex, cerebellum and thal-
amus. The different results can be explained by differences in
methodology. We restricted the analysis to cortical areas to
increase statistical power. Furthermore, we used the EMG as
reference signal and localized coherence changes (rather than co-
herence per se) in a single step procedure. Previous studies first
identified primary motor cortex as the source of maximum coher-
ence with the muscle. Subsequently, the authors searched for
sources coherent with primary motor cortex.
Oscillatory coupling and rest tremor Brain 2013: 136; 3659–3670 | 3667
at Un
iversitaets- u
nd
Lan
desb
iblio
thek
Du
esseldo
rf on
Feb
ruary
10
, 20
14
http
://brain
.ox
ford
jou
rnals.o
rg/
Do
wn
load
ed fro
m
Although the approach applied here is less sensitive than the
approach used previously, it provides improved reliability.
Association with tremor is not based on coherence peaks at
tremor frequency but on the contrast between tremor and
tremor-free epochs, allowing for the computation of group statis-
tics controlled for multiple comparisons. Moreover, the method
does not require removing the activity of strongly coherent
sources from the data prior to detection of weaker couplings
(Gross et al., 2001). This analysis step bears caveats, as incomplete
removal will lead to the detection of spurious coupling. Finally, it
accounts for the effect of volume conduction. Volume conduction,
also referred to as spatial leakage, results from suboptimal spatial
filtering and may substantially confound analysis of cortico-cortical
connectivity (Schoffelen and Gross, 2009; Palva and Palva, 2012).
Clinical relevance
The current study demonstrates a direct relationship between
subthalamic nucleus oscillations at tremor frequency and tremor
manifestation. Thus, subthalamic nucleus power and/or subthala-
mic nucleus-cortex coherence might potentially be used by closed-
loop DBS systems designed to suppress tremor. Subthalamic
nucleus power is a particularly promising parameter for triggering
DBS. In contrast to systems that use cortical action potentials as
triggers (Rosin et al., 2011), a system using subthalamic nucleus
power would not require additional cortical implants and would be
robust to slight changes in electrode position. Furthermore, online
computation of oscillatory power requires less computational
resources than other suggested control parameters, such as
phase-amplitude coupling (de Hemptinne et al., 2013).
This study provides important information on how subthalamic
nucleus power could be used by closed-loop systems. It suggests
that power increases at individual tremor frequency could serve as
a trigger signal. To achieve more robust tremor detection, we
propose to apply DBS whenever subthalamic nucleus power
increases at tremor frequency and its first harmonic and simultan-
eously decreases in the beta band.
ConclusionParkinsonian rest tremor is associated with an increase of cerebral
synchronization at tremor frequency and double tremor frequency.
The increase occurs in a network including subthalamic nucleus,
primary motor cortex, premotor cortex and posterior parietal
cortex. These results suggest the feasibility of tremor detection
based solely on cerebral oscillations.
AcknowledgementsThe authors would like to express their sincere gratefulness to the
patients who participated in this study. Furthermore, we are very
thankful to Medtronic for providing non-magnetic lead extensions.
In addition, we thank Mrs. E. Radisch for assistance with MRI
scans.
FundingThis work was supported by ERA-NET Neuron [Neuron-48-013 to
A.S.], by the Deutsche Forschungsgemeinschaft [FOR1328, SCHN
592/3-1 to A.S.] and by the European Commission [Marie Curie
Fellowship; FP7-PEOPLE-2009-IEF-253965 to M.B.].
Supplementary materialSupplementary material is available at Brain online.
ReferencesAmtage F, Henschel K, Schelter B, Vesper J, Timmer J, Lucking C, et al.
Tremor-correlated neuronal activity in the subthalamic nucleus of
parkinsonian patients. Neurosci Lett 2008; 442: 195–9.
Avants BB, Epstein CL, Grossman M, Gee JC. Symmetric diffeomorphic
image registration with cross-correlation: evaluating automated label-
ing of elderly and neurodegenerative brain. Med Image Anal 2008; 12:
26–41.
Benabid AL, Pollak P, Gervason C, Hoffmann D, Gao DM, Hommel M,
et al. Long-term suppression of tremor by chronic stimulation of the
ventral intermediate thalamic nucleus. Lancet 1991; 337: 403–6.
Bergman H, Wichmann T, DeLong MR. Reversal of experimental parkin-
sonism by lesions of the subthalamic nucleus. Science 1990; 249:
1436–8.
Bergman H, Wichmann T, Karmon B, DeLong MR. The primate subtha-
lamic nucleus. II. Neuronal activity in the MPTP model of parkinson-
ism. J Neurophysiol 1994; 72: 507–20.
Brittain J, Probert-Smith P, Aziz T, Brown P. Tremor suppression by
rhythmic transcranial current stimulation. Curr Biol 2013; 23: 436–40.
de Hemptinne C, Ryapolova-Webb ES, Air EL, Garcia PA, Miller KJ,
Ojemann JG, et al. Exaggerated phase–amplitude coupling in the pri-
mary motor cortex in Parkinson disease. Proc Natl Acad Sci USA 2013;
110: 4780–5.
Deiber MP, Pollak P, Passingham R, Landais P, Gervason C, Cinotti L,
et al. Thalamic stimulation and suppression of parkinsonian tremor.
Evidence of a cerebellar deactivation using positron emission tomog-
raphy. Brain 1993; 116 (Pt 1): 267–79.
Desmurget M, Turner RS. Motor aequences and the basal ganglia: kine-
matics, not habits. J Neurosci 2010; 30: 7685–90.
Deuschl G, Raethjen J, Baron R, Lindemann M, Wilms H, Krack P. The
pathophysiology of parkinsonian tremor: a review. J Neurol 2000; 247:
V33–48.
Elble RJ. Central mechanisms of tremor. J Clin Neurophysiol 1996; 13:
133–44.
Erbil N, Ungan P. Changes in the alpha and beta amplitudes of the
central EEG during the onset, continuation, and offset of long-duration
repetitive hand movements. Brain Res 2007; 1169: 44–56.
Florin E, Gross J, Reck C, Maarouf M, Schnitzler A, Sturm V, et al.
Causality between local field potentials of the subthalamic nucleus
and electromyograms of forearm muscles in Parkinson’s disease. Eur
J Neurosci 2010; 31: 491–8.
Forstmann BU, Keuken MC, Jahfari S, Bazin P-L, Neumann J, Schafer A,
et al. Cortico-subthalamic white matter tract strength predicts inter-
individual efficacy in stopping a motor response. Neuroimage 2012;
60: 370–5.
Goetz CG, Tilley BC, Shaftman SR, Stebbins GT, Fahn S, Martinez-
Martin P, et al. Movement Disorder Society-sponsored revision of
the Unified Parkinson’s Disease Rating Scale (MDS-UPDRS): scale
presentation and clinimetric testing results. Mov Disord 2008; 23:
2129–70.
3668 | Brain 2013: 136; 3659–3670 J. Hirschmann et al.
at Un
iversitaets- u
nd
Lan
desb
iblio
thek
Du
esseldo
rf on
Feb
ruary
10
, 20
14
http
://brain
.ox
ford
jou
rnals.o
rg/
Do
wn
load
ed fro
m
Gross J, Baillet S, Barnes GR, Henson R, Hillebrand A, Jensen O, et al.
Good practice for conducting and reporting MEG research.
Neuroimage 2013; 65: 349–63.
Gross J, Kujala J, Hamalainen M, Timmermann L, Schnitzler A,
Salmelin R. Dynamic imaging of coherent sources: studying neural
interactions in the human brain. Proc Natl Acad Sci USA 2001; 98:
694–9.
Hammond C, Bergman H, Brown P. Pathological synchronization in
Parkinson’s disease: networks, models and treatments. Trends
Neurosci 2007; 30: 357–64.
Heimer G, Rivlin-Etzion M, Bar-Gad I, Goldberg JA, Haber SN,
Bergman H. Dopamine replacement therapy does not restore the
full spectrum of normal pallidal activity in the 1-methyl-4-phenyl-
1,2,3,6-tetra-hydropyridine primate model of Parkinsonism.
J Neurosci 2006; 26: 8101–14.
Hellwig B, Haussler S, Lauk M, Guschlbauer B, Koster B, Kristeva-
Feige R, et al. Tremor-correlated cortical activity detected by electro-
encephalography. Clin Neurophysiol 2000; 111: 806–9.
Helmich RC, Hallett M, Deuschl G, Toni I, Bloem BR. Cerebral causes
and consequences of parkinsonian resting tremor: a tale of two
circuits? Brain 2012; 135: 3206–26.
Helmich RC, Janssen MJR, Oyen WJG, Bloem BR, Toni I. Pallidal
dysfunction drives a cerebellothalamic circuit into Parkinson tremor.
Ann Neurol 2011; 69: 269–81.
Hemm S, Coste J, Gabrillargues J, Ouchchane L, Sarry L, Caire F, et al.
Contact position analysis of deep brain stimulation electrodes on post-
operative CT images. Acta Neurochir 2009; 151: 823–9.
Hirschmann J, Ozkurt TE, Butz M, Homburger M, Elben S, Hartmann CJ,
et al. Distinct oscillatory STN-cortical loops revealed by simultaneous
MEG and local field potential recordings in patients with Parkinson’s
disease. Neuroimage 2011; 55: 1159–68.
Hirschmann J, Ozkurt TE, Butz M, Homburger M, Elben S, Hartmann CJ,
et al. Differential modulation of STN-cortical and cortico-muscular
coherence by movement and levodopa in Parkinson’s disease.
Neuroimage 2013; 68: 203–13.
Hoehn MM, Yahr MD. Parkinsonism: onset, progression and mortality.
Neurology 1967; 17: 427–42.
Hughes AJ, Daniel SE, Blankson S, Lees AJ. A clinicopathologic study of
100 cases of Parkinson’s disease. Arch Neurol 1993; 50: 140–8.
Hurtado JM, Rubchinsky LL, Sigvardt KA, Wheelock VL, Pappas CTE.
Temporal Evolution of Oscillations and Synchrony in GPi/Muscle
Pairs in Parkinson’s Disease. J Neurophysiol 2005; 93: 1569–84.
Hutchison WD, Lozano AM, Tasker RR, Lang AE, Dostrovsky JO.
Identification and characterization of neurons with tremor-
frequency activity in human globus pallidus. Exp Brain Res 1997;
113: 557–63.
Jenkinson M, Beckmann CF, Behrens TEJ, Woolrich MW, Smith SM. FSL.
Neuroimage 2012; 62: 782–90.
Krack P, Pollak P, Limousin P, Benazzouz A, Benabid AL. Stimulation of
subthalamic nucleus alleviates tremor in Parkinson’s disease. Lancet
1997; 350: 1675.
Kuhn AA, Kupsch A, Schneider G-H, Brown P. Reduction in subthalamic
8-35 Hz oscillatory activity correlates with clinical improvement in
Parkinson’s disease. Eur J Neurosci 2006; 23: 1956–60.
Lang AE, Lozano AM. Parkinson’s disease. Second of two parts. N Engl J
Med 1998; 339: 1130–43.
Lenz FA, Tasker RR, Kwan HC, Schnider S, Kwong R, Murayama Y, et al.
Single unit analysis of the human ventral thalamic nuclear group:
correlation of thalamic ‘tremor cells’ with the 3-6 Hz component of
parkinsonian tremor. J Neurosci 1988; 8: 754–64.
Levy R, Ashby P, Hutchison WD, Lang AE, Lozano AM, Dostrovsky JO.
Dependence of subthalamic nucleus oscillations on movement and
dopamine in Parkinson’s disease. Brain 2002; 125: 1196–209.
Litvak V, Eusebio A, Jha A, Oostenveld R, Barnes G, Foltynie T, et al.
Movement-related changes in local and long-range synchronization
in Parkinson’s disease revealed by simultaneous magnetoence-
phalography and intracranial recordings. J Neurosci 2012; 32:
10541–53.
Litvak V, Jha A, Eusebio A, Oostenveld R, Foltynie T, Limousin P, et al.
Resting oscillatory cortico-subthalamic connectivity in patients with
Parkinson’s disease. Brain 2011; 134: 359–74.
Liu X, Ford-Dunn H, Hayward G, Nandi D, Miall R, Aziz TZ, et al. The
oscillatory activity in the parkinsonian subthalamic nucleus investigated
using the macro-electrodes for deep brain stimulation. Clin
Neurophysiol 2002; 113: 1667–72.
Magarinos-Ascone CM, Figueiras-Mendez R, Riva-Meana C, Cordoba-
Fernandez A. Subthalamic neuron activity related to tremor
and movement in Parkinson’s disease. Eur J Neurosci 2000; 12:
2597–607.
Maris E, Oostenveld R. Nonparametric statistical testing of EEG- and
MEG-data. J Neurosci Methods 2007; 164: 177–90.
Mattout J, Henson RN, Friston KJ. Canonical source reconstruction for
MEG. Comput Intell Neurosci 2007; 2007: 1–10.
McAuley JH, Marsden CD. Physiological and pathological tremors and
rhythmic central motor control. Brain 2000; 123 (Pt 8): 1545–67.
Moran A, Bergman H, Israel Z, Bar-Gad I. Subthalamic nucleus functional
organization revealed by parkinsonian neuronal oscillations and syn-
chrony. Brain 2008; 131: 3395–409.
Mure H, Hirano S, Tang CC, Isaias IU, Antonini A, Ma Y, et al.
Parkinson’s disease tremor-related metabolic network: characteriza-
tion, progression, and treatment effects. Neuroimage 2011; 54:
1244–53.
Muthuraman M, Heute U, Arning K, Anwar AR, Elble RJ, Deuschl G,
et al. Oscillating central motor networks in pathological tremors and
voluntary movements. What makes the difference? NeuroImage 2012;
60: 1331–9.
Nolte G. The magnetic lead field theorem in the quasi-static approxima-
tion and its use for magnetoencephalography forward calculation in
realistic volume conductors. Phys Med Biol 2003; 48: 3637–52.
Nolte G, Bai O, Wheaton L, Mari Z, Vorbach S, Hallett M. Identifying
true brain interaction from EEG data using the imaginary part of co-
herency. Clin Neurophysiol 2004; 115: 2292–307.
Oliverira RM, Gurd JM, Nixon P, Marshall JC, Passingham RE.
Hypometria in Parkinson’s disease: automatic versus controlled pro-
cessing. Mov Disord 1998; 13: 422–7.
Oostenveld R, Fries P, Maris E, Schoffelen J-M. FieldTrip: open source
software for advanced analysis of MEG, EEG, and invasive electro-
physiological data. Comput Intell Neurosci 2011; 2011: 1–9.
Oswal A, Brown P, Litvak V. Movement related dynamics of subthalmo-
cortical alpha connectivity in Parkinson’s disease. Neuroimage 2013;
70: 132–42.
Ozkurt TE, Butz M, Homburger M, Elben S, Vesper J, Wojtecki L, et al.
High frequency oscillations in the subthalamic nucleus: a neurophysio-
logical marker of the motor state in Parkinson’s disease. Exp Neurol
2011; 229: 324–31.
Palva S, Palva JM. Discovering oscillatory interaction networks with M/
EEG: challenges and breakthroughs. Trends Cogn Sci (Regul Ed) 2012;
16: 219–30.
Pfurtscheller G, Graimann B, Huggins JE, Levine SP, Schuh LA.
Spatiotemporal patterns of beta desynchronization and gamma syn-
chronization in corticographic data during self-paced movement. Clin
Neurophysiol 2003; 114: 1226–36.
Pollok B, Gross J, Dirks M, Timmermann L, Schnitzler A. The cerebral
oscillatory network of voluntary tremor. J Physiol (Lond) 2004; 554:
871–8.
Pollok B, Makhloufi H, Butz M, Gross J, Chaieb L, Wojtecki L, et al.
Levodopa affects functional brain networks in parkinsonian resting
tremor. Mov Disord 2009; 24: 91–8.
Raz A, Vaadia E, Bergman H. Firing patterns and correlations of spon-
taneous discharge of pallidal neurons in the normal and the tremulous
1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine vervet model of parkin-
sonism. J Neurosci 2000; 20: 8559–71.
Reck C, Florin E, Wojtecki L, Krause H, Groiss S, Voges J, et al.
Characterisation of tremor-associated local field potentials in the sub-
thalamic nucleus in Parkinson’s disease. Eur J Neurosci 2009; 29:
599–612.
Oscillatory coupling and rest tremor Brain 2013: 136; 3659–3670 | 3669
at Un
iversitaets- u
nd
Lan
desb
iblio
thek
Du
esseldo
rf on
Feb
ruary
10
, 20
14
http
://brain
.ox
ford
jou
rnals.o
rg/
Do
wn
load
ed fro
m
Reck C, Himmel M, Florin E, Maarouf M, Sturm V, Wojtecki L, et al.
Coherence analysis of local field potentials in the subthalamic nucleus:
differences in parkinsonian rest and postural tremor. Eur J Neurosci
2010; 32: 1202–14.
Rosin B, Slovik M, Mitelman R, Rivlin-Etzion M, Haber SN, Israel Z, et al.
Closed-loop deep brain stimulation is superior in ameliorating parkin-
sonism. Neuron 2011; 72: 370–84.
Schnitzler A, Gross J. Normal and pathological oscillatory communication
in the brain. Nat Rev Neurosci 2005; 6: 285–96.
Schnitzler A, Munks C, Butz M, Timmermann L, Gross J. Synchronized
brain network associated with essential tremor as revealed by magne-
toencephalography. Mov Disord 2009; 24: 1629–35.
Schnitzler A, Timmermann L, Gross J. Physiological and pathological
oscillatory networks in the human motor system. J Physiol Paris
2006; 99: 3–7.
Schoffelen J-M, Gross J. Source connectivity analysis with MEG and EEG.
Hum Brain Mapp 2009; 30: 1857–65.
Sekihara K, Nagarajan SS, Poeppel D, Marantz A, Miyashita Y.
Application of an MEG eigenspace beamformer to reconstructing
spatio-temporal activities of neural sources. Hum Brain Mapp 2002;
15: 199–215.
Shimamoto SA, Ryapolova-Webb ES, Ostrem JL, Galifianakis NB,
Miller KJ, Starr PA. Subthalamic nucleus neurons are synchronized to
primary motor cortex local field potentials in Parkinson’s disease. J
Neurosci 2013; 33: 7220–33.
Stein JF, Aziz TZ. Does imbalance between basal ganglia and cerebellar
outputs cause movement disorders? Curr Opin Neurol 1999; 12:
667–9.
Taulu S, Simola J. Spatiotemporal signal space separation method for
rejecting nearby interference in MEG measurements. Phys Med Biol
2006; 51: 1759–68.
Thomson DJ. Spectrum estimation and harmonic analysis. Proc IEEE
1982; 70: 1055–96.
Timmermann L, Gross J, Dirks M, Volkmann J, Freund H-J, Schnitzler A.
The cerebral oscillatory network of parkinsonian resting tremor. Brain
2003; 126: 199–212.
Van Veen BD, Buckley KM. Beamforming: a versatile approach to spatial
filtering. IEEE ASSP Mag 1988; 5: 4–24.
Volkmann J, Joliot M, Mogilner A, Ioannides AA, Lado F, Fazzini E, et al.
Central motor loop oscillations in parkinsonian resting tremor revealed
magnetoencephalography. Neurology 1996; 46: 1359–9.
Wang S, Aziz TZ, Stein JF, Bain PG, Liu X. Physiological and harmonic
components in neural and muscular coherence in parkinsonian tremor.
Clin Neurophysiol 2006; 117: 1487–1498.
Wang S, Aziz TZ, Stein JF, Liu X. Time–frequency analysis of transient
neuromuscular events: dynamic changes in activity of the subthalamic
nucleus and forearm muscles related to the intermittent resting tremor.
J Neurosci Methods 2005; 145: 151–8.
Wang S, Chen Y, Ding M, Feng J, Stein JF, Aziz TZ, et al. Revealing the
dynamic causal interdependence between neural and muscular signals
in parkinsonian tremor. J Franklin Ins 2007; 344: 180–195.
Zirh TA, Lenz FA, Reich SG, Dougherty PM. Patterns of bursting occur-
ring in thalamic cells during parkinsonian tremor. Neuroscience 1998;
83: 107–21.
3670 | Brain 2013: 136; 3659–3670 J. Hirschmann et al.
at Un
iversitaets- u
nd
Lan
desb
iblio
thek
Du
esseldo
rf on
Feb
ruary
10
, 20
14
http
://brain
.ox
ford
jou
rnals.o
rg/
Do
wn
load
ed fro
m
Rest tremor and oscillatory coupling – supplementary material
Fig. S1: Time course of power before and after tremor onset.
The figure shows the dynamics of group average EMG (upper row), MEG (middle row) and LFP power
(lower row). Black horizontal lines indicate mean baseline power (-9 to 0 s) and shaded areas depict
the standard error of the mean. Left column: Power at individual tremor frequency ± 0.5 Hz. Right
column: Power at 20 Hz ± 5 Hz. Please note that LFP power at tremor frequency started to fluctuate
around a higher mean value after tremor onset. LFP beta power decreased between -4 to 0 s,
followed by a transient increase at tremor onset and another transient decrease. MEG beta power
had similar dynamics, but the power decrease intensified gradually as tremor continued.
Rest tremor and oscillatory coupling – supplementary material
Fig. S2: Tremor-related changes in cortico-muscular coherence occurred in spatially separated
cortical areas.
Cortical areas that showed significant increases in cortico-muscular coherence during tremor are
displayed (p < 0.05; N = 8). All of these areas belonged to a single cluster, i.e. they were
interconnected in space by super-threshold voxels. However, distinct local maxima within that
cluster were clearly distinguishable. Cross hairs mark the regions of interest from which time series
were reconstructed. M1 = primary motor cortex; PPC = parietal cortex; PMC = premotor cortex.
CHAPTER SIX
Magnetoencephalographyand Neuromodulation
Alfons Schnitzler*,†,1, Jan Hirschmann*,†
*Institute of Clinical Neuroscience and Medical Psychology, Medical Faculty, Heinrich-Heine-UniversityDusseldorf, Dusseldorf, Germany†Department of Neurology, University Hospital Dusseldorf, Dusseldorf, Germany1Corresponding author: e-mail address: [email protected]
Contents
1. Introduction 121
2. MEG and TMS 122
2.1 MEG informs TMS 122
2.2 MEG characterizes TMS effects 124
3. MEG and Transcranial Direct or Alternate Current Stimulation 125
4. MEG and DBS 125
4.1 Simultaneous MEG and local field potential recordings in PD patients 126
4.2 MEG characterizes DBS effects 130
5. Conclusions and Outlook 132
References 133
Abstract
Magnetoencephalography (MEG) is a noninvasive method which allows recordings
of human brain activity with excellent temporal and good spatial resolution. In this
chapter, we review applications of MEG in neuromodulation. We provide an overview
of studies which used MEG to optimize parameters for neuromodulation and to charac-
terize the electrophysiological effects of brain stimulation. In particular, we discuss how
MEG may be employed to study deep brain stimulation. In this context, we describe
the problems arising from stimulation artifacts and present approaches to solve them.
1. INTRODUCTION
Magnetoencephalography (MEG) is one of the most widely used
techniques tomeasure brain activity noninvasively in humans. Based on sup-
erconducting interference device technology, MEG is able to detect the ex-
tremely small magnetic fields resulting from joint activity of several thousand
neurons (Hamalainen, Hari, Ilmoniemi, Knuutila, & Lounasmaa, 1993).
International Review of Neurobiology, Volume 107 # 2012 Elsevier Inc.ISSN 0074-7742 All rights reserved.http://dx.doi.org/10.1016/B978-0-12-404706-8.00007-3
121
Modern MEG systems are equipped with hundreds of sensors which sample
brain activity with a frequency of several thousand Hertz. Due to recent ad-
vances in source reconstruction methodology, task-related or spontaneous
activity can be localized with sub-centimeter spatial resolution (Barnes,
Hillebrand, Fawcett, & Singh, 2004). Thus, MEG has excellent temporal
and good spatial resolution and is therefore ideally suited to study the effects
of neuromodulation in humans.
MEG provides helpful information at all stages of neuromodulation re-
search. It may be used (i) to identify the anatomical targets and the appro-
priate timing of brain stimulation, (ii) to characterize immediate and lasting
neurophysiological effects of intervention, and (iii) to acquire basic knowl-
edge about the pathophysiological mechanisms which make therapeutic
neuromodulation necessary. In this chapter, we will review recent applica-
tions ofMEG in neuromodulation. In the first part, wewill give an overview
of studies combining MEG with transcranial magnetic stimulation (TMS).
In the second part, we will provide a short outlook on prospective applica-
tions of MEG in combination with transcranial current stimulation. Finally,
we will discuss in detail how MEG studies may contribute to the under-
standing and improvement of deep brain stimulation (DBS) and give exam-
ples of studies combining these two techniques.
2. MEG AND TMS
TMS is a noninvasive brain stimulation technique which is used
to study casual relationships between brain activity and behavior (Pascual-
Leone, Bartres-Faz, & Keenan, 1999). Moreover, TMS is applied for
therapeutic purposes (Barr et al., 2011; Croarkin, Wall, & Lee, 2011;
Corti, Patten, & Triggs, 2012). In TMS, a current is produced in a coil
which is positioned above the target brain area. The current induces a
strong and focal magnetic field which in turn induces a secondary current
in the targeted brain area. The induced current interferes with local
processing and may result in an observable change in behavior.
2.1. MEG informs TMS
Naturally, the outcome of TMS depends on the specific target. A study by
Raij et al. (2008) showed howMEGmay be employed to obtain knowledge
about the appropriate stimulation target. The authors measured
somatosensory-evoked fields (SEFs) by means of MEG and identified indi-
vidual SEF latencies and the underlying generator sources. Subsequently,
they combined median nerve stimulation, electroencephalography (EEG)
122 Alfons Schnitzler and Jan Hirschmann
and TMS and let each somatosensory stimulus be followed by a single TMS
pulse. The pulse was targeted at the region functionally identified as second-
ary somatosensory cortex (S2) in the previous MEG experiment. Subjects
were instructed to react to each somatosensory stimulus as fast as possible.
The authors found that a TMS pulse delivered 15–40 ms after somatosen-
sory stimulation decreased reaction times and shifted the 140 ms component
of the evoked potential forward in time, suggesting that S1!S2 reciprocal
pathways were facilitated by TMS of S2.
Another example of howMEGmay be used to inform TMS protocols is
given by a recent study by Thut et al. (2011). In this study, MEG was
recorded to obtain information on brain target area and frequency of stim-
ulation. The authors recorded the MEG of subjects performing a visual at-
tention task. Using frequency analysis and MEG source reconstruction
techniques, they identified the alpha frequency and parietal brain region
which showed the strongest modulation by attention in each individual sub-
ject (Fig 6.1A). In a subsequent session, they applied TMS at the individual
alpha frequency over the individual alpha generator in a combined
Map1
(90-°)
-2.5
-2
-1.5
-1
-0.5
0
A B
0.5
Am
plit
ude [m
V]
Tms1 Tms2 Tms3 Tms4 Tms5
- PO4
- CP4
1
1.5
2
2.5
Map1
(270-°)
Map5
(270-°)
Map5
(90-°)
Figure 6.1 MEG-informed rhythmic TMS entrains endogenous alpha oscillations. (A)
Beamformer reconstruction of the alpha source showing the strongest modulation
by attention in theMEG experiment. The source was localized using group average data.
(B) Sensor waveforms from EEG electrodes CP4 (red) and PO4 (black) evoked by each of
five TMS pulses (TMS 1–5) delivered at individual alpha frequency. Heat maps show the
topography of evoked activity at 90 !C (upper row) and 270 !C (lower row) of the alpha
cycle. From Thut et al. (2011).
123Magnetoencephalography and Neuromodulation
EEG–TMS experimental setup. They were able to show that on-going al-
pha oscillations can be externally entrained by rhythmic TMS (Fig 6.1B).
This result is of importance as it points toward a new approach in
neuromodulation which aims at modulating endogenous, on-going oscilla-
tory activity.
The two studies reviewed above nicely demonstrate the advantages and
limitations of MEG in combination with stimulation. Both studies used
MEG to obtain individually adjusted stimulation parameters but then ap-
plied EEG to measure the immediate electrophysiological effects of TMS.
In contrast to EEG, MEG allows high-density measurements with little
preparation time, making MEG an optimal tool to perform quick localizer
experiments. Moreover, MEG is often preferred over EEG for source local-
ization because its localization accuracy depends to a lesser extent on the ac-
curacy and complexity of the forward model (Leahy, Mosher, Spencer,
Huang, & Lewine, 1998). However, the greatest strength of MEG is also
its greatest drawback when used in combination with TMS. Simultaneous
MEG–TMS recordings are rendered impossible by the extreme sensitivity of
MEG to magnetic fields. Characterization of immediate TMS effects must
therefore be performed by less sensitive devices. Combined EEG–TMS re-
cordings have lately emerged as a powerful tool for this purpose (Thut &
Miniussi, 2009).
2.2. MEG characterizes TMS effects
As discussed in the previous section, MEG is not suitable to evaluate the im-
mediate neurophysiological effects of TMS. However, the lasting effects of
repetitive TMS (rTMS) can be characterized by means of MEG. Depending
on the frequency of stimulation, rTMS either decreases (1 Hz) or increases
( 5 Hz) cortical excitability (Siebner & Rothwell, 2003). The effect can
outlast stimulation by several minutes (Peinemann et al., 2004). While mo-
tor evoked potentials are the standard output measure of excitability in the
TMS literature, MEG studies have additionally investigated modulations of
oscillatory activity. Tamura et al. (2005) found the postmovement rebound
of motor cortical beta oscillations to be reduced after 1 Hz rTMS. Another
study made use of intermittent theta burst TMS (iTBS), an excitability en-
hancing protocol (Hsu et al., 2011). Complementary to the results by
Tamura et al., the study reported iTBS to increase postmovement beta syn-
chronization. The same approach was used to study therapeutic effects of
TMS. Lorenz, Muller, Schlee, Langguth, and Weisz (2010) recorded the
124 Alfons Schnitzler and Jan Hirschmann
MEG of tinnitus patients after treatment by rTMS. They found the auditory
steady state response to be reduced along with perceived tinnitus loudness.
3. MEG AND TRANSCRANIAL DIRECT OR ALTERNATE
CURRENT STIMULATION
Characterization of lasting effects of neuromodulation by MEG is not
limited to TMS. For example, MEG has been used to investigate the
neurophysiological after-effects of transcutaneous electrical nerve stimula-
tion (Hoshiyama & Kakigi, 2000; Murakami et al., 2010). Moreover,
the approach will likely play an important role in unveiling the
mechanisms by which both transcranial direct current stimulation (tDCS)
and transcranial alternating current stimulation (tACS) modulate
behavior. In particular, it may help to clarify whether and to what extent
tACS affects endogenous oscillations. Currently, there is both evidence
for (Marshall, Helgadottir, Molle, & Born, 2006; Kirov, Weiss, Siebner,
Born, & Marshall, 2009; Zaehle, Rach, & Herrmann, 2010) and against
(Antal et al., 2008) the notion that tACS can modulate brain oscillations
to an extent that it affects behavior. In the future, tACS may be used
with individually tailored parameters which may be provided by MEG
recordings. This procedure may enable researchers to not only impose a
rhythm on the stimulated areas but also to selectively enhance or reduce
specific components of on-going oscillatory activity, much like rhythmic
TMS (Thut et al., 2011). In this form, tACS might become a powerful
therapeutic tool for the treatment of disorders which are characterized by
abnormal synchronization, such as schizophrenia, epilepsy, or Parkinson’s
disease (PD; Schnitzler & Gross, 2005).
4. MEG AND DBS
DBS is an invasive electrical brain stimulation intervention routinely
applied to treat symptoms of PD, essential tremor, dystonia, and several pain
syndromes (Perlmutter &Mink, 2006). Further, it is being tested as a poten-
tial treatment for a variety of other disorders, such as depression, obsessive
compulsive disorder, and Tourette syndrome. In DBS, macroelectrodes are
implanted into the target area. The electrodes deliver current pulses gener-
ated by a subcutaneously implanted stimulator. Target area and stimulation
parameters depend on the disorder which is treated. For treatment of PD,
125Magnetoencephalography and Neuromodulation
the most common target is the subthalamic nucleus (STN), which is typi-
cally stimulated with a frequency of 130 Hz.
Despite intensive research, the mechanisms of action by which DBS
alleviates PD symptoms are not finally understood. Invasive recordings in
animal models and humans have identified inhibitory and excitatory effects
of DBS as well as complex polysynaptic responses (Kringelbach, Jenkinson,
Owen, & Aziz, 2007). Given that the clinical effect of DBS is similar to the
effect of leasioning the STN (Bergman, Wichmann, & DeLong, 1990), it is
often hypothesized that DBS acts by disrupting pathological activity. In fact,
there is growing evidence that PD symptoms are correlated with or caused
by pathological oscillatory activity which can be recorded in the STN and
other basal ganglia nuclei (Brown, 2003). Further, it was shown that
symptom-associated synchronization is not restricted to the basal ganglia
but can be found within a wider network of cortical and subcortical areas
(Timmermann et al., 2003; Pollok et al., 2008). Not surprisingly,
growing interest in synchronized networks has motivated MEG studies
addressing cortical synchronization with basal ganglia oscillations and
combination of MEG and DBS.
4.1. Simultaneous MEG and local field potential recordings inPD patients
Much has been learned about PD pathophysiology by studying local field
potentials (LFPs) recorded by DBS electrodes. It was revealed that STN ac-
tivity of PD patients is characterized by strong beta oscillations (13–35 Hz)
which are modulated by movement and dopaminergic medication (Brown
et al., 2001; Kuhn et al., 2004; Priori et al., 2004). Beta oscillations are
phase–amplitude coupled to high frequency oscillations (>200 Hz) which
are themselves dopamine responsive (Lopez-Azcarate et al., 2010; Ozkurt
et al., 2011). Despite their substantial contribution to PD research, the
insights gained through LFP recordings are limited. DBS electrodes
record from only a small fraction of the spatially distributed motor
network. In order to gain insights into network connectivity, LFP
recordings need to be combined with multisensor recording techniques,
such as whole head MEG or high-density EEG.
4.1.1 STN–cortical coupling
Several studies investigated STN–cortical coupling in PD patients selected
for therapeutic DBS. Most of them analyzed coherence, a frequency-
domain measure of similarity. Coherence measures the degree of linear
126 Alfons Schnitzler and Jan Hirschmann
dependency between two signals. It quantifies to what degree two signals
have a stable phase and amplitude relation over time. Coherence may arise
when membrane potential fluctuations in two distant neuronal populations
A and B are coordinated in time, for example, such that incoming action
potentials from A reach B at the time of maximum excitability (Fries,
2005). In this case, population A would have maximum gain on population
B.Note, however, that coherence analysismay yield the very same valuewhen
timing is such that action potentials from A reach B at the time of minimum
excitability, as it merely captures the presence of any coordination in time.
EEG studies investigating coupling between STN LFPs and cortical ac-
tivity showed that beta coherence is reduced by movement (Cassidy et al.,
2002; Lalo et al., 2008) and found indications for a modulation by
dopaminergic medication (Williams et al., 2002). Further, it was
demonstrated that STN–cortical coherence has a frequency-dependent
topography on the sensor level (Fogelson et al., 2006). However, the
limited number of EEG channels precluded localization of coherence on
the source level. Recently, localization of coherent sources was achieved
by simultaneous recordings of LFPs and MEG (Hirschmann et al., 2011;
Litvak et al., 2011). In two independent studies, Hirschmann et al. and
Litvak et al. recorded PD patients at rest after withdrawal of dopaminergic
medication and independently reported the same frequency-dependent
spatial distribution of coherence (Fig. 6.2). STN–cortical beta coherence
was found in medial sensorimotor and premotor cortex ipsilateral to the
recorded STN. In contrast, alpha coherence localized to areas in ipsilateral
temporal cortex and brainstem. In contrast to alpha coherence with
temporal cortex, beta coherence with sensorimotor cortex was restricted to
one or two bipolar contacts of the DBS electrode, suggesting a focal origin
of coherent beta oscillations within the recording area of the electrode
(Hirschmann et al., 2011). In summary, these studies revealed two distinct
functional connections involving the STN which operate in distinct
frequency bands. The results may hint at a general mechanism in
communication between distant brain areas: Simultaneously on-going
interactions involving the same anatomical structures might be distinguished
by frequency.
Litvak et al. extended their analysis of MEG–LFP resting state coherence
beyond the description of its spatial distribution (Litvak et al., 2011, 2012).
They investigated causal interactions between STN and cortex and found
motor cortex to drive STN in the beta band, in agreement with previous
studies (Williams et al., 2002; Lalo et al., 2008). Thus, enhanced STN
127Magnetoencephalography and Neuromodulation
beta activity may have its origin in the cortex. In fact, high frequency optical
stimulation of STN afferents from primary motor cortex (M1) was shown to
effectively ameliorate motor symptoms in the rat model of PD (Gradinaru,
Mogri, Thompson, Henderson, & Deisseroth, 2009), demonstrating the
importance of STN–cortical pathways in PD pathophysiology. However,
20 Hz stimulation of STN afferents did not worsen PD symptoms,
A
C
B
D
10
5
10
5
Figure 6.2 The spatial distribution of STN–cortical coherence is frequency dependent.
Upper row: Distribution of spatial maxima of significantly coherent sources. (A) Spatial
maxima of alpha coherent sources from a group of eight bilateral implanted Parkinson's
diseases patients. (B) Spatial maxima of beta coherent sources. Sources coherent to the
left STN have been mirrored across the midsagittal plane. Colors code electrode contact
pairs recording local field potentials, that is, the reference signal for coherence compu-
tations. Blue¼channel 01 (most ventral), green¼channel 12, orange¼channel 23, and
red¼channel 03. From Hirschmann et al. (2011). Lower row: Statistical parametric maps
(SPMs) showing the relative distribution of alpha and beta STN–cortical coherence. (C)
SPM showing regions where alpha coherence is significantly higher than beta coher-
ence. (D) SPM showing regions where beta coherence is significantly higher than alpha
coherence. Color bars indicate t-statistic. From Litvak et al. (2011).
128 Alfons Schnitzler and Jan Hirschmann
suggesting that motor cortical input at beta frequencies may not be the
cause of symptoms. On the contrary, a recent MEG–LFP study in PD
patients found M1–STN beta coherence to be anticorrelated with
akinesia, meaning that the highest coherence was found in patients with
the least akinesia (Hirschmann et al., 2012). These results suggest that
M1–STN beta coherence may be of physiological nature or serve
compensation, rather than reflecting or causing slowing of movement.
In combination, the studies summarized above indicate that the STN
receives rhythmic input at beta frequencies from motor cortex. This
input may not be pathological per se, but may be unusually amplified in
the basal ganglia of PD patients, possibly due to increased circuit
resonance resulting from dopamine depletion (Eusebio et al., 2009).
4.1.2 Modulation of STN–cortical coupling
Tobetter understand the role of STN–cortical coupling in normal and impaired
motor function, it is important to study its modulation by dopaminergic med-
ication and voluntary movement. Recent research indicates that these two fac-
tors strongly interact.While administration of levodopawas reported to have no
effect on STN–cortical coherence at rest (Litvak et al., 2011), it was found to
modulate coupling during movement (Hirschmann et al., 2012; Litvak et al.,
2012). Litvak et al. investigated the effect of levodopa administration on
movement-related changes in M1–STN–cortical coherence (Litvak et al.,
2012). They reported a short-lived increase of gamma band coherence at
movement onset, which was intensified by administration of dopaminergic
medication. The effect of medication was positively correlated with
improvement of PD motor symptoms. Hirschmann et al. measured baseline
levels of M1–STN coherence while subjects performed two different motor
tasks and found coherence to be suppressed by levodopa in the beta band
(Hirschmann et al., 2012). Thus, levodopa administration may specifically
affect movement-related oscillatory coupling. Moreover, the studies discussed
above provide further evidence for the antagonistic relationship between
movement-related beta and gamma oscillations (Brown, 2003): Gamma
activity is enhanced by movement and dopaminergic medication while beta
activity is reduced.
4.1.3 Concluding remarks on simultaneous MEG and LFP recordings
The results reviewed above demonstrate how simultaneous LFP–MEG re-
cordings can extend our knowledge about basic motor system neurophysiol-
ogy and its pathological alterations. The interplay between DBS surgery and
129Magnetoencephalography and Neuromodulation
MEG recordings is a good example of how basic research may profit from
applied neuromodulation. In turn, neuromodulation relies on basic research.
For example, LFP–MEG coherence analysis provides information on the
functional connectivity of the DBS target. This information may help to ex-
plain some of the effects and side-effects of DBS and might contribute to the
identification of new and better stimulation targets.
4.2. MEG characterizes DBS effects
In the studies discussed above, MEG has been combined with
recordings from DBS electrodes. Other studies aimed at investigating the
immediate neurophysiological effects of DBS and applied high frequency
stimulation while recording MEG. As the magnetic fields induced by
DBS may exceed physiological magnetic fields by several orders of magni-
tude, this experimental approach requires tools for artifact suppression. So
far, two methods have been employed: beamforming and temporal signal
space separation (tSSS).
4.2.1 DBS artifact suppression by beamforming
The term beamforming refers to the application of a spatial filter which
lets activity from a selected location pass while blocking all other signals
(Van Veen & Buckley, 1988). Over the past few years, beamforming has
become an established method for source reconstruction in MEG research
(Hillebrand, Singh, Holliday, Furlong, & Barnes, 2005). As it is designed
to cancel interfering signals, beamforming provides implicit protection
against artifacts. This property has previously been exploited to suppress
artifacts which can arise in simultaneous LFP–MEG recordings (Litvak
et al., 2010).
Early MEG–DBS studies made use of a beamforming variant called syn-
thetic aperture magnetometry to study the effects of DBS in a small number
of chronic pain patients (Kringelbach et al., 2007; Ray et al., 2007, 2009).
These studies demonstrated the feasibility of measuring MEG during DBS
and reported modulations of power in areas involved in pain sensation.
Recently, the authors of this pioneering work refined their source
reconstruction methodology by modifying the beamformer formulation
to obtain a so-called null-beamformer (Mohseni et al., 2012). In null-
beamforming, the constraints on spatial filtering are extended by the
requirement to force output to zero for signals originating from a selected
location (Van Veen & Buckley, 1988). Based on the observation that the
entry points of DBS electrodes into the skull are a source of artifacts, the
130 Alfons Schnitzler and Jan Hirschmann
authors chose to suppress signals originating from burr holes. Using the null-
beamformer, they analyzed data from a chronic pain patient implanted with
electrodes for stimulation of the anterior cingulate cortex (ACC). They
found the null-beamformer to be superior to conventional beamforming
when localizing the source of high frequency stimulation for evaluation
purposes (Fig. 6.3A). Further, they reported that DBS decreased theta
power in caudal ACC, an area involved in pain sensation. Notably, the
stimulation-induced power decrease in ACC was less strong when the
experiment was repeated after 12 months, suggesting that chronic DBS
triggered plasticity changes.
4.2.2 DBS artifact suppression by tSSS
Other studies have employed tSSS to suppress DBS artifacts. In contrast to
beamforming, tSSS is a not a source reconstruction method but a
preprocessing procedure serving artifact suppression. tSSS is the temporal
extension of signal space separation, a commercial algorithmwhich separates
100 fT/cm
100 ms
A B
Figure 6.3 Methods for artifact suppression in MEG–DBS recordings. (A) The
null-beamformer (lower row) provides better localization of a known source than a con-
ventional beamformer (upper row) in MEG–DBS experiments. The red dot marks the
location of the DBS electrode contact producing the 130 Hz activity which was localized
for evaluation purposes. Contours indicate 130 Hz power. From Mohseni et al. (2012).
(B) Auditory-evoked fields (AEFs) in two MEG sensors recorded during DBS ON (blue)
and DBS OFF (red). The upper row shows AEFs before application of tSSS. The lower
row shows AEFs in the same sensors after application of tSSS. From Airaksinen et al.
(2011).
131Magnetoencephalography and Neuromodulation
the MEG signal into two subspaces: one for the sources inside the sensor
array and one for interfering sources outside the array (Taulu & Kajola,
2005; Taulu & Simola, 2006). When the signal is reconstructed the latter
are left out, resulting in suppression of external artifacts (Fig. 6.3B).
Makela, Taulu, Pohjola, Ahonen, and Pekkonen (2007) first applied
tSSS to analyze the effects of STN stimulation on MEG activity in a PD pa-
tient. They reported that DBS suppressed spontaneous sensorimotor activity
in the 3–10 Hz range. This pilot experiment was followed by two group
studies investigating the effects of DBS in PD patients (Airaksinen et al.,
2011, Airaksinen et al., 2012). In the first study, subjects received
auditory and somatosensory stimulations during DBS OFF and DBS ON
(Airaksinen et al., 2011). It was found that DBS increased auditory-
evoked fields ipsilateral to auditory stimulation. In the second study, the
authors analyzed changes in oscillatory activity and correlated the latter to
clinical ratings of PD motor symptom severity (Airaksinen et al., 2012).
They reported that alpha and beta power in pericentral areas correlated
positively with rigidity scores only when DBS was on, suggesting that DBS
affects the behavioral relevance of oscillatory activity in sensorimotor
cortex. However, patients were medicated in this study, so that the
observed effects could not be attributed uniquely to DBS.
5. CONCLUSIONS AND OUTLOOK
We have provided several examples of how MEG may be used to in-
form and to evaluate different brain stimulation techniques, such as TMS or
DBS. We have elaborated on the strengths of MEG but also discussed the
artifacts which complicate combinations of MEG and stimulation. The fu-
ture role of MEG in neuromodulation research will depend to a large extent
on the development of methods for artifact suppression. Fortunately, much
progress has been made. Algorithms such as tSSS and null-beamforming
provide a promising basis for future research.
Provided that artifacts are handled efficiently, one of the greatest contri-
butions of MEG to neuromodulation research will continue to be the anal-
ysis of stimulation effects on source level network activity. Although it is not
trivial to reconstruct numerous simultaneously active sources based onMEG
sensor data, MEG remains the best available, noninvasive method for inves-
tigation of synchronized networks in humans. In the future, network studies
addressing coupling between all voxels in the brain will become more fre-
quent and will provide important insights into large-scale connectivity
132 Alfons Schnitzler and Jan Hirschmann
patterns (Schoffelen & Gross, 2011; Hillebrand, Barnes, Bosboom,
Berendse, & Stam, 2012; Hipp, Hawellek, Corbetta, Siegel, & Engel,
2012). As a large body of evidence supports the importance of
synchronized networks in normal and pathological brain function, it must
be expected that effective brain stimulation, be it invasive or noninvasive,
modifies these networks. Future studies will uncover such modifications
and thereby extend our knowledge about the mechanisms of action of
targeted brain stimulation. Most likely, MEG studies will make a significant
contribution.
REFERENCESAiraksinen, K., Butorina, A., Pekkonen, E., Nurminen, J., Taulu, S., Ahonen, A., et al.
(2012). Somatomotor mu rhythm amplitude correlates with rigidity during deep brainstimulation in Parkinsonian patients. Clinical Neurophysiology, 123, 2010–2017.
Airaksinen, K., Makela, J. P., Taulu, S., Ahonen, A., Nurminen, J., Schnitzler, A., et al.(2011). Effects of DBS on auditory and somatosensory processing in Parkinson’s disease.Human Brain Mapping, 32, 1091–1099.
Antal, A., Boros, K., Poreisz, C., Chaieb, L., Terney, D., & Paulus, W. (2008). Compara-tively weak after-effects of transcranial alternating current stimulation (tACS) on corticalexcitability in humans. Brain Stimulation, 1, 97–105.
Barnes, G. R., Hillebrand, A., Fawcett, I. P., & Singh, K. D. (2004). Realistic spatial samplingfor MEG beamformer images. Human Brain Mapping, 23, 120–127.
Barr, M. S., Farzan, F., Wing, V. C., George, T. P., Fitzgerald, P. B., & Daskalakis, Z. J.(2011). Repetitive transcranial magnetic stimulation and drug addiction. International Re-view of Psychiatry, 23, 454–466.
Bergman, H., Wichmann, T., & DeLong, M. R. (1990). Reversal of experimental parkin-sonism by lesions of the subthalamic nucleus. Science, 249, 1436–1438.
Brown, P. (2003). Oscillatory nature of human basal ganglia activity: Relationship to thepathophysiology of Parkinson’s disease. Movement Disorders, 18, 357–363.
Brown, P., Oliviero, A., Mazzone, P., Insola, A., Tonali, P., & Di Lazzaro, V. (2001). Do-pamine dependency of oscillations between subthalamic nucleus and pallidum inParkinson’s disease. The Journal of Neuroscience, 21, 1033–1038.
Cassidy, M., Mazzone, P., Oliviero, A., Insola, A., Tonali, P., Di Lazzaro, V., et al. (2002).Movement-related changes in synchronization in the human basal ganglia. Brain, 125,1235–1246.
Corti, M., Patten, C., & Triggs, W. (2012). Repetitive transcranial magnetic stimulation ofmotor cortex after stroke: A focused review. American Journal of Physical Medicine & Re-habilitation, 91, 254–270.
Croarkin, P. E., Wall, C. A., & Lee, J. (2011). Applications of transcranial magnetic stimu-lation (TMS) in child and adolescent psychiatry. International Review of Psychiatry, 23,445–453.
Eusebio, A., Pogosyan, A., Wang, S., Averbeck, B., Gaynor, L. D., Cantiniaux, S., et al.(2009). Resonance in subthalamo-cortical circuits in Parkinson’s disease. Brain, 132,2139–2150.
Fogelson, N., Williams, D., Tijssen, M., van Bruggen, G., Speelman, H., & Brown, P.(2006). Different functional loops between cerebral cortex and the subthalmic area inParkinson’s disease. Cerebral Cortex, 16, 64–75.
133Magnetoencephalography and Neuromodulation
Fries, P. (2005). A mechanism for cognitive dynamics: Neuronal communication throughneuronal coherence. Trends in Cognitive Sciences, 9, 474–480.
Gradinaru, V., Mogri, M., Thompson, K. R., Henderson, J. M., & Deisseroth, K. (2009).Optical deconstruction of Parkinsonian neural circuitry. Science, 324, 354–359.
Hamalainen, M., Hari, R., Ilmoniemi, R. J., Knuutila, J., & Lounasmaa, O. V. (1993).Magnetoencephalography—Theory, instrumentation, and applications to noninvasivestudies of the working human brain. Reviews of Modern Physics, 65, 413–497.
Hillebrand, A., Barnes, G. R., Bosboom, J. L., Berendse, H. W., & Stam, C. J. (2012). Fre-quency-dependent functional connectivity within resting-state networks: An atlas-basedMEG beamformer solution. NeuroImage, 59, 3909–3921.
Hillebrand, A., Singh, K. D., Holliday, I. E., Furlong, P. L., & Barnes, G. R. (2005). A newapproach to neuroimaging with magnetoencephalography. Human Brain Mapping, 25,199–211.
Hipp, J. F., Hawellek, D. J., Corbetta, M., Siegel, M., & Engel, A. K. (2012). Large-scalecortical correlation structure of spontaneous oscillatory activity. Nature Neuroscience,15, 884–890.
Hirschmann, J., Ozkurt, T. E., Butz, M., Homburger, M., Elben, S., Hartmann, C. J., et al.(2011). Distinct oscillatory STN-cortical loops revealed by simultaneous MEG and localfield potential recordings in patients with Parkinson’s disease.NeuroImage, 55, 1159–1168.
Hirschmann, J., Ozkurt, T. E., Butz, M., Homburger, M., Elben, S., Hartmann, C. J., et al.(2012). Differential modulation of STN-cortical and cortico-muscular coherence bymovement and levodopa in Parkinson’s disease. In Press.
Hoshiyama, M., & Kakigi, R. (2000). After-effect of transcutaneous electrical nerve stimu-lation (TENS) on pain-related evoked potentials and magnetic fields in normal subjects.Clinical Neurophysiology, 111, 717–724.
Hsu, Y.-F., Liao, K.-K., Lee, P.-L., Tsai, Y.-A., Yeh, C.-L., Lai, K.-L., et al. (2011). Inter-mittent theta burst stimulation over primary motor cortex enhances movement-relatedbeta synchronisation. Clinical Neurophysiology, 122, 2260–2267.
Kirov, R., Weiss, C., Siebner, H. R., Born, J., & Marshall, L. (2009). Slow oscillation elec-trical brain stimulation during waking promotes EEG theta activity and memoryencoding. PNAS, 106, 15460–15465.
Kringelbach, M. L., Jenkinson, N., Green, A. L., Owen, S. L. F., Hansen, P. C.,Cornelissen, P. L., et al. (2007). Deep brain stimulation for chronic pain investigatedwith magnetoencephalography. Neuroreport, 18, 223–228.
Kringelbach, M. L., Jenkinson, N., Owen, S. L. F., & Aziz, T. Z. (2007). Translational prin-ciples of deep brain stimulation. Nature Reviews. Neuroscience, 8, 623–635.
Kuhn, A. A., Williams, D., Kupsch, A., Limousin, P., Hariz, M., Schneider, G.-H., et al.(2004). Event-related beta desynchronization in human subthalamic nucleus correlateswith motor performance. Brain, 127, 735–746.
Lalo, E., Thobois, S., Sharott, A., Polo, G., Mertens, P., Pogosyan, A., et al. (2008). Patternsof bidirectional communication between cortex and basal ganglia during movement inpatients with Parkinson disease. The Journal of Neuroscience, 28, 3008–3016.
Leahy, R., Mosher, J., Spencer, M., Huang, M., & Lewine, J. (1998). A study of dipolelocalization accuracy for MEG and EEG using a human skull phantom. Electroencephalog-raphy and Clinical Neurophysiology, 107, 159–173.
Litvak, V., Eusebio, A., Jha, A., Oostenveld, R., Barnes, G., Foltynie, T., et al. (2012).Movement-related changes in local and long-range synchronization in Parkinson’s dis-ease revealed by simultaneous magnetoencephalography and intracranial recordings. TheJournal of Neuroscience, 32, 10541–10553.
Litvak, V., Eusebio, A., Jha, A., Oostenveld, R., Barnes, G. R., Penny, W. D., et al. (2010).Optimized beamforming for simultaneous MEG and intracranial local field potential re-cordings in deep brain stimulation patients. NeuroImage, 50, 1578–1588.
134 Alfons Schnitzler and Jan Hirschmann
Litvak, V., Jha, A., Eusebio, A., Oostenveld, R., Foltynie, T., Limousin, P., et al. (2011).Resting oscillatory cortico-subthalamic connectivity in patients with Parkinson’s disease.Brain, 134, 359–374.
Lopez-Azcarate, J., Tainta, M., Rodrıguez-Oroz, M. C., Valencia, M., Gonzalez, R.,Guridi, J., et al. (2010). Coupling between beta and high-frequency activity in the hu-man subthalamic nucleus may be a pathophysiological mechanism in Parkinson’s disease.The Journal of Neuroscience, 30, 6667–6677.
Lorenz, I., Muller, N., Schlee, W., Langguth, B., & Weisz, N. (2010). Short-term effects ofsingle repetitive TMS sessions on auditory evoked activity in patients with chronic tin-nitus. Journal of Neurophysiology, 104, 1497–1505.
Makela, J. P., Taulu, S., Pohjola, J., Ahonen, A., & Pekkonen, E. (2007). Effects of sub-thalamic nucleus stimulation on spontaneous sensorimotor MEG activity in a Parkinso-nian patient. International Congress Series, 1300, 345–348.
Marshall, L., Helgadottir, H., Molle, M., & Born, J. (2006). Boosting slow oscillations duringsleep potentiates memory. Nature, 444, 610–613.
Mohseni, H. R., Smith, P. P., Parsons, C. E., Young, K. S., Hyam, J. A., Stein, A., et al.(2012). MEG can map short and long-term changes in brain activity following deep brainstimulation for chronic pain. PloS One, 7, e37993.
Murakami, T., Takino, R., Ozaki, I., Kimura, T., Iguchi, Y., &Hashimoto, I. (2010). High-frequency transcutaneous electrical nerve stimulation (TENS) differentially modulatessensorimotor cortices: An MEG study. Clinical Neurophysiology, 121, 939–944.
Ozkurt, T. E., Butz, M., Homburger, M., Elben, S., Vesper, J., Wojtecki, L., et al. (2011).High frequency oscillations in the subthalamic nucleus: A neurophysiological marker ofthe motor state in Parkinson’s disease. Experimental Neurology, 229, 324–331.
Pascual-Leone, A., Bartres-Faz, D., & Keenan, J. P. (1999). Transcranial magnetic stimula-tion: Studying the brain-behaviour relationship by induction of “virtual lesions” Philo-sophical Transactions of the Royal Society of London. Series B, Biological Sciences, 354,1229–1238.
Peinemann, A., Reimer, B., Loer, C., Quartarone, A., Munchau, A., Conrad, B., et al.(2004). Long-lasting increase in corticospinal excitability after 1800 pulses of subthresh-old 5 Hz repetitive TMS to the primary motor cortex. Clinical Neurophysiology, 115,1519–1526.
Perlmutter, J. S., &Mink, J.W. (2006). Deep brain stimulation.Annual Review of Neuroscience,29, 229–257.
Pollok, B., Makhloufi, H., Butz, M., Gross, J., Timmermann, L., Wojtecki, L., et al. (2008).Levodopa affects functional brain networks in Parkinsonian resting tremor. MovementDisorders, 24, 91–98.
Priori, A., Foffani, G., Pesenti, A., Tamma, F., Bianchi, A. M., Pellegrini, M., et al. (2004).Rhythm-specific pharmacological modulation of subthalamic activity in Parkinson’s dis-ease. Experimental Neurology, 189, 369–379.
Raij, T., Karhu, J., Kicic, D., Lioumis, P., Julkunen, P., Lin, F.-H., et al. (2008). Parallelinput makes the brain run faster. NeuroImage, 40, 1792–1797.
Ray, N. J., Jenkinson, N., Kringelbach, M. L., Hansen, P. C., Pereira, E. A., Brittain, J. S.,et al. (2009). Abnormal thalamocortical dynamics may be altered by deep brain stimu-lation: Using magnetoencephalography to study phantom limb pain. Journal of ClinicalNeuroscience, 16, 32–36.
Ray, N., Kringelbach, M., Jenkinson, N., Owen, S., Davies, P., Wang, S., et al. (2007).Using magnetoencephalography to investigate brain activity during high frequency deepbrain stimulation in a cluster headache patient. Biomedical Imaging and Intervention Journal,3, e25.
Schnitzler, A., & Gross, J. (2005). Normal and pathological oscillatory communication in thebrain. Nature Reviews. Neuroscience, 6, 285–296.
135Magnetoencephalography and Neuromodulation
Schoffelen, J.-M., & Gross, J. (2011). Improving the interpretability of all-to-all pairwisesource connectivity analysis in MEG with nonhomogeneous smoothing. Human BrainMapping, 32, 426–437.
Siebner, H., & Rothwell, J. (2003). Transcranial magnetic stimulation: New insights intorepresentational cortical plasticity. Experimental Brain Research, 148, 1–16.
Tamura, Y., Hoshiyama, M., Nakata, H., Hiroe, N., Inui, K., Kaneoke, Y., et al. (2005).Functional relationship between human rolandic oscillations and motor cortical excit-ability: An MEG study. The European Journal of Neuroscience, 21, 2555–2562.
Taulu, S., & Kajola, M. (2005). Presentation of electromagnetic multichannel data: Thesignal space separation method. Journal of Applied Physics, 97, 124905–124910.
Taulu, S., & Simola, J. (2006). Spatiotemporal signal space separation method for rejectingnearby interference in MEG measurements. Physics in Medicine and Biology, 51,1759–1768.
Thut, G., & Miniussi, C. (2009). New insights into rhythmic brain activity from TMS–EEGstudies. Trends in Cognitive Sciences, 13, 182–189.
Thut, G., Veniero, D., Romei, V., Miniussi, C., Schyns, P., & Gross, J. (2011). RhythmicTMS causes local entrainment of natural oscillatory signatures. Current Biology, 21,1176–1185.
Timmermann, L., Gross, J., Dirks, M., Volkmann, J., Freund, H.-J., & Schnitzler, A. (2003).The cerebral oscillatory network of parkinsonian resting tremor. Brain, 126, 199–212.
Van Veen, B. D., & Buckley, K. M. (1988). Beamforming: A versatile approach to spatialfiltering. IEEE ASSP Magazine, 5, 4–24.
Williams, D., Tijssen, M., Van Bruggen, G., Bosch, A., Insola, A., Di Lazzaro, V., et al.(2002). Dopamine-dependent changes in the functional connectivity between basalganglia and cerebral cortex in humans. Brain, 125, 1558–1569.
Zaehle, T., Rach, S., & Herrmann, C. S. (2010). Transcranial alternating current stimulationenhances individual alpha activity in human EEG. PloS One, 5, e13766.
136 Alfons Schnitzler and Jan Hirschmann